September 22, 2026

What happened

An Optimum on-site U.S. Analyst Threat Intelligence role involves collecting and analyzing intelligence on threats, threat actors, malware campaigns, and relevant tactics, techniques, and procedures. [1]

The role includes mapping adversary behavior to MITRE ATT&CK, producing actionable reports and alerts, supporting incident response and threat hunting, and maintaining threat intelligence platforms. [2]

The article also lists an Application Security Specialist position at SMBC Group in Ireland with a hybrid work arrangement. [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Analyst Threat Intelligence Optimum | USA | On-site – View job details As an Analyst Threat Intelligence, you will collect and analyze intelligence to identify threats, threat actors, malware campaigns, and relevant TTPs.

2. [2]

   You will map adversary behavior to MITRE ATT&CK, produce actionable reports and alerts, support incident response and threat hunting, and maintain threat intelligence platforms.

3. [3]

   Application Security Specialist SMBC Group | Ireland | Hybrid – View job details As an Application Security Specialist, … More → The post Cybersecurity jobs available right now: September 22, 2026 appeared first on Help Net Security .

Read the original article →

What happened

IDC forecasts that European organizations will spend nearly $470 billion on AI in 2030, with spending growing at a 35% compound annual rate from 2025. [1]

At that projected growth rate, the European AI market would more than quadruple over five years. [2]

Generative AI is projected to account for 55.4% of total European AI spending by 2030. [3]

The article identifies agentic AI as the main driver of the projected spending growth. [4]

Why it matters

Companies are moving from single-purpose copilots toward multiple agents working together while operating under the EU AI Act. [5]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   European organizations will spend nearly $470 billion on AI in 2030, IDC forecasts, with spending growing at a compound annual rate of 35% from 2025.

2. [2]

   At that rate, the market more than quadruples in five years.

3. [3]

   Generative AI will account for 55.4% of the total by 2030.

4. [4]

   agentic AI is the main driver.

5. [5]

   Companies are moving from single-purpose copilots to several agents working together, and they are doing it under the EU AI Act, whose … More → The post European AI spending is on track to reach nearly $470 billion by 2030 appeared first on Help Net Security .

Read the original article →

What happened

By the second half of 2026, national competent authorities across the EU are actively reviewing NIS2 compliance documentation. [1]

Under Article 20(1), senior management at essential and important entities can be held personally liable for infringements. [2]

Why it matters

The source characterizes potential personal liability as a factor that tends to concentrate executive attention. [2]

ENISA’s 2025 NIS Investments report found that 34% of EU organizations report severe skills shortages in IAM implementation. [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   By the second half of 2026, national competent authorities across the EU are actively reviewing NIS2 compliance documentation.

2. [2]

   Under Article 20(1) of the directive, senior management at essential and important entities can be held personally liable for infringements — a detail that tends to concentrate executive attention.

3. [3]

   ENISA’s 2025 NIS Investments report found that 34% of EU organizations report severe skills shortages specifically in identity and access management (IAM) implementation.

Read the original article →

What happened

CISA added a now-patched vulnerability affecting Zyxel GS1900 series switches to its Known Exploited Vulnerabilities catalog, citing evidence of active exploitation. [1]

The vulnerability is identified as CVE-2026-7273 and has a CVSS score of 8.8. [2]

CVE-2026-7273 is described as a stack-based buffer overflow vulnerability that could result in arbitrary operating… [2]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Cybersecurity and Infrastructure Security Agency (CISA) on Monday added a now-patched security flaw impacting Zyxel GS1900 series switches to its Known Exploited Vulnerabilities (KEV) catalog, citing evidence of active exploitation.

2. [2]

   The vulnerability, tracked as CVE-2026-7273 (CVSS score: 8.8), is a stack-based buffer overflow vulnerability that could result in arbitrary operating

Read the original article →

What happened

Researchers built 5G-Shark, a low-cost tool that lures phones to a fake base station and questions them, then used it to audit commercial 5G networks. [1]

On the tested standalone-5G networks, operators correctly concealed phones’ permanent identities in all but one case. [2]

The tested networks nevertheless issued temporary identifiers in a pattern predictable enough for an observer to track 5G subscribers. [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Researchers from the i2CAT Foundation, the University of Murcia, and NEC Laboratories Europe built a low-cost tool called 5G-Shark that lures a target phone onto a fake base station and questions it, then used it to audit commercial 5G networks.

2. [2]

   On the standalone-5G networks they tested, operators concealed the phone’s permanent identity correctly in every case but one.

3. [3]

   The same networks still handed out temporary IDs in a pattern predictable enough that an observer can … More → The post A cheap fake base station can still track 5G subscribers appeared first on Help Net Security .

Read the original article →

What happened

Security researcher Patrick Wardle demonstrated in a proof of concept that malware already running on a Mac could take over Meta’s Muse assistant and use the broad access granted to the app. [1]

The demonstrated technique changes a hidden setting so dictated microphone prompts are sent to the attacker instead of Meta. [2]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Malware already running on a Mac can quietly take over Meta's Muse assistant and use the broad access its owner granted the app, security researcher Patrick Wardle has shown in a proof-of-concept released on September 21.

2. [2]

   It works by changing a hidden setting so that when the user taps the microphone and dictates a prompt, the words go to the attacker instead of Meta.

Read the original article →

What happened

The North Korea-linked WaterPlum group uses fake job interviews to target freelance developers and blockchain or Web3 specialists, infecting at least 30,000 devices across more than 100 countries. [1] [2] [3] [4] [5]

Actors impersonate AI, cryptocurrency, or NFT companies and direct candidates to download files for coding tests or supposed bug fixes; the downloads contain malware families including BeaverTail, InvisibleFerret, OtterCookie, OtterCandy, and StoatWaffle. [4] [5] [6]

The article reports that Japan’s authorities shut down the country’s first known laptop farm and that the FBI continues prosecuting facilitators who help North Korean IT workers operate through such arrangements. [15] [16]

Why it matters

The campaign reportedly stole funds or credentials from more than 7,000 cryptocurrency wallets and transferred 1.7 billion JPY in cryptocurrency assets to North Korea. [2] [7] [8]

A successful infection can enable theft of browser authentication data, keystrokes, screenshots, wallet data, files, and shared-folder contents, as well as espionage, intellectual-property theft, and lateral movement into an employer’s environment. [9] [10] [11]

The advisory links WaterPlum operators and North Korean IT workers to the same organization and reports shared IP addresses involving laptop farms, freelance platforms, and a crypto-exchange job application. [12] [13] [14]

Known limitations

The advisory says the described techniques are only examples and that the actors continuously evolve and refine their methods. [17]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   North Korea-linked WaterPlum runs the Contagious Interview campaign, infecting over 30,000 devices using a fake job interview.

2. [2]

   At least 30,000 devices infected across more than 100 countries, funds or credentials stolen from over 7,000 cryptocurrency wallets, and a total of 1.7 billion yen, roughly $10.71 million, funneled back to North Korea.

3. [3]

   The victims are almost always the same type: freelance developers and specialists in blockchain and Web3 work.

4. [4]

   WaterPlum actors pose as recruiters or hiring managers, often impersonating real AI, crypto, or NFT companies, and reach out on social media, job boards, and freelance platforms.

5. [5]

   Candidates get invited to a technical interview, then told to download a file to complete a coding test or fix a supposed bug on the video call.

6. [6]

   The advisory names five malware families riding inside these downloads, BeaverTail , InvisibleFerret , OtterCookie , OtterCandy, and StoatWaffle , each doing its own piece of the job: one steals browser credentials, another opens a backdoor, another sets up a remote access trojan to move deeper into the machine.

7. [7]

   “WaterPlum actors have infected at least 30,000 devices in more than 100 countries and exfiltrated funds or account credentials from over 7,000 cryptocurrency wallets.

8. [8]

   WaterPlum actors have transferred 1.7 billion Japanese yen (JPY) (equivalent to 10.71 million USD) of cryptocurrency assets to the Democratic People’s Republic of Korea (DPRK).” That file is never what it claims to be.

9. [9]

   The report notes that a successful infection can give WaterPlum a way into the victim’s employer, opening the door to espionage, stolen source code, and further movement inside a company’s systems.

10. [10]

    “Beyond immediate credential theft, successful infections provide WaterPlum actors opportunities to infiltrate organizations employing targeted developers, enabling espionage, intellectual property theft, and additional lateral movement in corporate environments.

11. [11]

    “Other sensitive data targeted for exfiltration includes: Authentication data stored in web browsers (ID, password, etc.); Clipboard information, key-logs (recorded keystrokes), screenshots; Cryptocurrency-wallet data (private key, seed phrase, etc.); and Any files or data of interest to the actors on a PC or in shared folders (ID pictures of driver’s licenses, passports, etc.)” Here’s a simpler and more natural version, while keeping the same details: Japan says this is no longer just a theoretical threat.

12. [12]

    Japan’s NPA and the FBI assess that both WaterPlum operators and some North Korean IT workers report to the same place, the 313 General Bureau of the Munitions Industry Department, under the Workers’ Party’s Central Committee.

13. [13]

    The advisory draws one more thread together: WaterPlum’s cyberattack operators and North Korea’s IT worker network aren’t running in parallel.

14. [14]

    They’ve been caught using the same IP addresses to access laptop farms, register on freelance platforms, and apply for that same crypto exchange job.

15. [15]

    Authorities have already shut down the country’s first known “laptop farm,” run by a local facilitator who had moved several hundred million yen in cryptocurrency out of Japan.

16. [16]

    In the US, the FBI says it continues to prosecute people who help North Korean IT workers operate this way, across several states and under different charges.

17. [17]

    “The techniques described in this advisory are only examples; actors continuously evolve and refine their methods.” concludes the advisory.

Read the original article →

What happened

SideCopy has been observed using spear-phishing lures against academic institutions in India, expanding its strategic focus beyond government entities. [1]

Trellix researchers state that SideCopy operations typically begin with spear-phishing campaigns abusing mshta.exe to execute malicious scripts and circumvent standard security protocols. [2]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   The threat actor known as SideCopy has been observed using spear-phishing lures to target academic institutions in India, expanding their strategic focus beyond government entities.

2. [2]

   "SideCopy campaign operations typically initiate through spear-phishing campaigns that leverage the abuse of mshta.exe to execute malicious scripts and circumvent standard security protocols," Trellix researchers

Read the original article →

What happened

Scammers are using a $249 website toolkit to market unverified AI subscriptions costing up to $2,000 annually, with genuine Google sign-in screens making the sites appear convincing. [1]

Malwarebytes found more than 100 websites built with the same toolkit and linked by closely related developer details. [2]

The network includes sites copying names of existing products, including GPT-6 Astra, DaVinci Resolve, PixAI, and OpenCut. [3]

Why it matters

Interpretation: The combination of copied product identities and genuine Google sign-in screens may make fraudulent subscription sites harder for users to distinguish from legitimate services. [1] [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Scammers are using a $249 website toolkit to sell unverified AI subscriptions worth up to $2,000 a year, and a genuine Google sign-in screen is what makes the sites convincing.

2. [2]

   Malwarebytes found more than 100 websites built this way, all tied to the same toolkit and closely related developer details.

3. [3]

   The network includes sites that copy the names of existing products, among them GPT-6 Astra, DaVinci Resolve, PixAI, and OpenCut.

Read the original article →

What happened

A malicious npm package named "indexed-btree" was observed hiding malicious behavior in application code rather than using lifecycle scripts. [1]

Checkmarx said indexed-btree mimicked the legitimate sorted-btree package, described as an ordinary B-tree/indexing utility. [2]

Why it matters

The observed tactic was characterized as indicating that threat actors are likely shifting tactics in response to recent security controls. [1]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A malicious npm package named "indexed-btree" has been observed hiding its malicious behavior within application code rather than using lifecycle scripts, indicating that threat actors are likely shifting tactics in response to recent security controls.

2. [2]

   "Indexed-btree is a malicious npm package mimicking the legit sorted-btree package, an ordinary B-tree/indexing utility," Checkmarx said.

Read the original article →

What happened

Hacktron researchers used two chained vulnerabilities—a crafted-image flaw in Discourse’s libheif processing and an OpenAI SSO flaw—to reach an internal OpenAI software repository through an employee’s connected Codex and GitHub access. [1] [2] [3] [4] [5] [6]

The image-processing vulnerability could let an attacker gain control of a Discourse server, while the SSO flaw exposed ChatGPT and Codex accounts belonging to people logged into OpenAI’s forum. [3] [7] [8]

The researchers said skilled human guidance remained important and characterized the activity as not completely autonomous hacking. [18] [19]

OpenAI fixed its portion of the problem about 14 hours after the initial report and later paid a $6,500 bounty; the researchers said they avoided sensitive information and used a harmless pull request to demonstrate access. [24] [25] [26] [27] [28]

Why it matters

Using Claude Opus 5, the researchers built a working exploit overnight after an earlier Claude model failed to produce a reliable exploit against default Discourse; they then used it against OpenAI’s forum. [9] [10] [11] [12] [13]

Claude initially refused to write an exploit for a remote system, but the researchers obtained its assistance by presenting the task as a capture-the-flag exercise; the agent took over the test forum server within four hours. [14] [15] [16] [17]

Hacktron spent less than $3,000 in AI tokens over two months and reported related weaknesses affecting services or software from Slack, Meta, and GitHub; adapting the exploit to a new company took one or two days on average. [20] [21] [22] [23]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Using Anthropic’s Claude, the researchers went from investigating an image-processing flaw to accessing an internal OpenAI software repository in less than 72 hours.

2. [2]

   To get inside OpenAI, researchers Harsh Jaiswal, Mohan Pedhapati, and Rahul Maini found two vulnerabilities and chained them together.

3. [3]

   Uploading a specially crafted image could trigger a flaw in the library, allowing an attacker to gain control of the Discourse server.

4. [4]

   This was a flaw in OpenAI’s single sign-on (SSO) system, which lets people access one service using an account from another.

5. [5]

   It wasn’t just public users that had signed into this system; OpenAI employees were signed in, too, allowing Hacktron to access one employee’s account.

6. [6]

   This gave the researchers access to OpenAI’s internal software repository.

7. [7]

   It involved a bug in an image-upload feature in the Discourse community forum software.

8. [8]

   The SSO flaw gave Hacktron access to the ChatGPT and Codex accounts of people who had logged in to OpenAI’s Discourse forum.

9. [9]

   It originally used Opus 4.8, one of Anthropic’s recent Claude models, to discover the issue in libheif.

10. [10]

    But it couldn’t use the model to build a reliable exploit that worked against the default version of Discourse.

11. [11]

    Using that model, the researchers were able to build a working exploit overnight.

12. [12]

    A task that Opus 4.8 had failed to complete across several sessions was solved by Opus 5 within hours of its release.

13. [13]

    The researchers then used the resulting exploit against OpenAI’s forum.

14. [14]

    The second interesting aspect is that the researchers had to fool Claude into helping them do it.

15. [15]

    The model refused to write an exploit for a remote system because it considered the request unethical.

16. [16]

    So the researchers had to present the task as a capture-the-flag exercise (a hacking competition) to get it to play ball.

17. [17]

    When they did that, the agent took over the test forum server within four hours.

18. [18]

    People have been able to use ready-made hacking tools for years without really understanding how they work, but it took real expertise to pick through software, find hidden flaws, and turn them into reliable exploits.

19. [19]

    Hacktron says skilled human guidance was still important and that this was not completely autonomous hacking.

20. [20]

    Hacktron spent less than $3,000 in AI tokens during its two-month research project, which involved three researchers and uncovered vulnerabilities affecting several major companies.

21. [21]

    The image-processing bug became the basis of a wider project called HEIF Heist .

22. [22]

    Hacktron found related security weaknesses affecting services and software from companies including Slack, Meta, and GitHub.

23. [23]

    Tweaking the HEIF exploit to target a new company took a day or two on average.

24. [24]

    They deliberately avoided viewing sensitive information.

25. [25]

    The researchers didn’t do anything damaging with that access.

26. [26]

    So they instructed the employee’s Codex account to create a harmless pull request—a proposed software change—in an internal OpenAI repository.

27. [27]

    OpenAI fixed its part of the problem about 14 hours after Hacktron submitted its initial report.

28. [28]

    It later paid the researchers a $6,500 bounty for the OpenAI-side flaw.

Read the original article →

What happened

SpyCloud analyzed 10,000 EPA-registered water and wastewater organizations and found 1,787 with active infostealer exposure, meaning identity data from stolen credentials was exposed. [1] [2]

Among the 1,787 organizations with active infostealer exposure, 258 carried credentials for operational-technology or remote-access systems. [9]

Why it matters

In one reported case, a single infected device at an unnamed smart-meter technology provider contained saved logins linked to about 167 U.S. utility-metering tenants, creating a reported cascading supply-chain exposure. [3] [4] [5]

SpyCloud’s investigations officer said infostealer logs commonly include session cookies, credentials and autofill data, which can enable hijacked authenticated sessions and access to corporate email or VPNs without an alert. [6] [7] [8]

Known limitations

The research did not focus on OT devices running critical utility processes, and the report said its exposure findings should not be interpreted as exposure of specific OT devices. [10] [11]

The report said larger operators and the vendor supply chain were overrepresented, small utilities were underrepresented, and the findings measure identity exposure rather than confirmed intrusion. [12] [13] [14]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   water and wastewater organizations have identity data actively exposed from infostealers harvesting their credentials, according to research published Tuesday.

2. [2]

   The company built a database of 66,845 Environmental Protection Agency-registered systems, examined internet domains and ultimately analyzed 10,000 organizations, finding that 1,787 showed active infostealer exposure.

3. [3]

   In one case, a single infected device at a smart meter technology provider that SpyCloud didn’t name contained saved logins linked to roughly 167 different U.S.

4. [4]

   utility metering tenants — meaning that one exposure opened the door to many more.

5. [5]

   That “cascading supply chain exposure” was one of the biggest findings of the report that SpyCloud shared exclusively with CyberScoop, said Jason Lancaster, chief investigations officer at the cyber firm, along with the quantitative approach” to measure exposure overall.

6. [6]

   “There’s examples here and there, but that was a standout example of, here’s a tangible thing that is an exposure right now.” Generally, “Infostealer exposure means the attacker isn’t guessing anymore, they’ve got legitimate points of entry,” Lancaster said.

7. [7]

   “In the investigations I’ve worked, that log data usually contains stolen session cookies, credentials, and autofill info pulled straight off the infected device.

8. [8]

   That’s enough to walk right past [multifactor authentication] by hijacking an already-authenticated session, log into corporate email or VPNs without raising a single alert, and sit there quietly for weeks while they map out the network.” Still, the study had limitations.

9. [9]

   It did, however, find that 258 of the 1,787 organizations with active infostealer exposure carried credentials to operational technology or remote-access systems.

10. [10]

    It doesn’t address what apparently led to the cyberattacks that unfolded in Minnesota and elsewhere this summer: internet-exposed programmable logic controllers .

11. [11]

    It’s “important to note that our research did not focus on OT devices which run the most critical processes within these utilities, and any exposure we cite herein should not be interpreted as exposure of specific OT devices,” the report said.

12. [12]

    Some of the report’s conclusions were about the limitations of the data itself.

13. [13]

    “Exposure concentrated in larger operators and in the vendor supply chain; small utilities were largely underrepresented,” SpyCloud said.

14. [14]

    “That’s a pattern in the data, not a claim that every water system nationwide carries this risk — and it measures identity exposure, not confirmed intrusion.” It’s the first study of its kind that SpyCloud has done for a specific industry, so the company doesn’t have comparisons to other industries, Lancaster said.

Read the original article →

What happened

Akamai reports that verified AI crawlers, including ChatGPT, have shifted from reading web pages to sending high-frequency POST requests. [1]

In Akamai’s 30-day analysis of global customers, ecommerce represented 44.8% of AI-bot POST transactions, while travel reached 30% in one month. [2]

Why it matters

The article distinguishes GET requests, which ask for a page, from POST requests, which tell a site to perform an action such as logging in or adding an item to a cart. [3] [4]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Akamai has watched verified AI crawlers, ChatGPT among them, move from reading web pages to sending high-frequency POST requests.

2. [2]

   In a 30-day analysis of its global customers, ecommerce accounted for 44.8% of those AI bot POST transactions, and travel climbed to 30% in a single month.

3. [3]

   A GET request asks a website for a page.

4. [4]

   A POST request tells the site to do something, like log a user in, add an item to a cart, … More → The post Somewhere in your traffic logs, a bot is doing more than looking appeared first on Help Net Security .

Read the original article →

What happened

GreyNoise reported that a Chinese-speaking threat actor exploited CVE-2026-7273 in unpatched ZyXEL GS1900 Smart Managed Switches and exfiltrated sensitive data from 996 devices across 48 countries. [1]

Why it matters

The affected switches were predominantly located in Italy, the United States, Taiwan, South Korea, and several European Union countries. [2]

Known limitations

The supplied evidence does not state the exploitation prerequisites, affected firmware versions, mitigation steps, or whether a permanent fix is available. [1] [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A Chinese-speaking threat actor has exploited a vulnerability (CVE-2026-7273) in unpatched ZyXEL GS1900 Smart Managed Switches and has exfiltrated sensitive data from 996 devices across 48 countries, GreyNoise reported on Monday.

2. [2]

   The affected switches are predominantly located in Italy, the US, Taiwan, South Korea, and a number of EU countries.

3. [3]

   CVE-2026-7273 exploitation is part of an unfolding operation The Zyxel GS1900 Series is a line of Gigabit Ethernet switches aimed at small and mid-sized business … More → The post Attacker compromised nearly 1000 Zyxel switches since August (CVE-2026-7273) appeared first on Help Net Security .

Read the original article →

What happened

The article reports that Muse can connect to services and receive macOS permissions for protected resources, including files, microphone, camera, location, and calendars. [1] [2]

According to the reported finding, locally running code or a terminal command could change an undocumented Muse setting controlling its dictation-transcription server. [3]

Redirecting dictation traffic to an attacker-controlled server could expose voice prompts and the authentication token for the victim’s Muse account. [4]

Why it matters

The reported issue is not remote code execution against a clean Mac: exploitation first requires local code execution, such as malware, a malicious application, or social engineering. [5] [6]

The article says a compromised AI agent could consolidate access to multiple services and operating-system permissions behind one authenticated interface. [7] [8]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Meta says Muse can handle appointments, forms, customer-service interactions, purchases, document creation, and connections to services such as WhatsApp, email, calendars, and social platforms.

2. [2]

   It may also receive macOS permissions to access protected resources, including files, the microphone, camera, location, and calendars.

3. [3]

   According to Ars Technica , Wardle found that a locally running application or terminal command could alter an undocumented Muse configuration setting that controls the server used for dictation transcription.

4. [4]

   By redirecting dictation traffic to an attacker-controlled server, an attacker could capture voice prompts and obtain the authentication token for the victim’s Muse account.

5. [5]

   This is not a remote-code-execution vulnerability that can compromise an otherwise clean Mac.

6. [6]

   The attacker first needs a way to run code locally, such as through malware, a malicious application, or social engineering.

7. [7]

   PROTECT YOUR IDENTITY Traditional infostealer malware must independently locate browser data, credentials, documents, chat histories, and other valuable material.

8. [8]

   A compromised AI agent could lower that barrier by bundling access to multiple services and operating-system permissions behind one already authenticated interface.

Read the original article →

What happened

The source reports that GPT-6 Astra broke the Enigma message Nr. 172, “MVUEH,” after analyzing unbroken messages and using the repeated place name “ROSENOW ROSENOW” as a crib. [1] [2] [3] [4] [5]

The reported result included developing Python and C++ software for an Enigma simulator and Bombe, then finding the correct key and plaintext for the MVUEH message. [5]

Why it matters

The source characterizes the break as having been carried out entirely by GPT-6 Astra, with Carter Leffer directing it to test whether it could break an unbroken message. [1] [6]

Known limitations

The source says the authors are still analyzing GPT-6 Astra’s logs to determine exactly how it executed the break. [7]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Carter Leffer only directed GPT­6 Astra to see if it could break any of the unbroken Enigma messages published on the Crypto Cellar Research web page.

2. [2]

   After analysing the unbroken messages on the website, it decided that the most promising message was Nr.

3. [3]

   172, MVUEH and it also quickly suspected that the plaintext of Nr.

4. [4]

   After trying many different approaches, GPT­6 Astra focused on using the repeated place name ROSENOW ROSENOW as a crib.

5. [5]

   After developing the necessary Python and C++ software for an Enigma simulator and an Enigma Bombe, GPT­6 Astra started a thorough break with the ROSENOW crib, which in the end resulted in the correct key and plaintext for the MVUEH message being found.

6. [6]

   This is pretty amazing: However, the most astonishing thing about this break is that the GPT­6 Astra did it entirely on its own.

7. [7]

   We are still analysing the GPT­6 Astra logs to see exactly how it executed the break.

Read the original article →

What happened

Viettel Cyber Security researcher Dinh Ho Anh Khoa reported that CVE-2026-65660, initially classified by Microsoft as a spoofing flaw with a 6.5 CVSS score, enables authenticated remote code execution. [1]

CVE-2026-65660 affects SharePoint Server 2016, SharePoint Server 2019, and SharePoint Server Subscription Edition. [2]

Known limitations

The supplied evidence truncates the statement about patches, so patch availability or status cannot be determined. [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A SharePoint Server vulnerability that Microsoft initially classified as a spoofing flaw with a CVSS score of 6.5 actually enables authenticated remote code execution, according to full technical details published today by Viettel Cyber Security researcher Dinh Ho Anh Khoa.

2. [2]

   The flaw, CVE-2026-65660, affects SharePoint Server 2016, 2019, and Subscription Edition.

3. [3]

   Patches have been

Read the original article →

What happened

A Linux kernel KVM virtualization flaw affecting ARM64 hosts with nested virtualization enabled can expose freed host memory to a guest virtual machine. [1]

Tracked as CVE-2026-89775, the flaw allows a guest to read and write host kernel memory. [2]

Why it matters

The discovering researcher says the flaw can enable a guest to escape and run code on the host machine. [2]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A new flaw in the Linux kernel's KVM virtualization code for ARM64 processors can leave a freed piece of host memory exposed to a guest virtual machine on hosts with nested virtualization enabled.

2. [2]

   The bug, tracked as CVE-2026-89775, allows a guest to read and write host kernel memory, and the researcher who found it says it can be used to escape the guest and run code on the host machine.

Read the original article →

What happened

Gartner reported that 41% of surveyed CISOs experienced at least one deepfake-related social-engineering incident during an employee audio call in the prior 12 months. [1]

Gartner reported that 36% of surveyed CISOs experienced at least one deepfake-related social-engineering incident during a video call in the prior 12 months. [2]

Why it matters

Among surveyed CISOs, 79% reported at least one phishing, spear-phishing, or business-email-compromise incident in the prior 12 months. [3]

Among surveyed CISOs, 58% reported at least one vishing or smishing incident in the prior 12 months. [4]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   41% of CISOs reported at least one social engineering incident involving a deepfake during an employee audio call in the past 12 months, according to Gartner.

2. [2]

   36% reported the same for a video call.

3. [3]

   79% of CISOs surveyed reported at least one phishing, spear-phishing, or business email compromise incident in the past 12 months.

4. [4]

   58% reported a vishing or smishing incident.

Read the original article →

What happened

Three Elsevier domains or web portals redirected users to a page branded “LAPSUS$ GROUP, Chapter II,” which included a signed statement taunting the FBI and a countdown to a future victim. [1]

Cloudskope researchers reported that the redirect lasted at least 78 minutes, from approximately 7:49 p.m. CT until it was cleared before 10:09 p.m. CT on September 21, 2026. [2]

Known limitations

The supplied evidence states that Elsevier had not yet offered an explanation for how the redirect occurred. [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Three domains / web portals belonging to Dutch academic publishing company Elsevier have been redirecting users to a page branded “LAPSUS$ GROUP, Chapter II,” carrying a signed statement that taunted the FBI and counted down to a future victim.

2. [2]

   According to Cloudskope researchers, the redirect ran for at least 78 minutes, from roughly 7:49pm CT until it was cleared before 10:09pm CT on September 21, 2026.

3. [3]

   Elsevier is yet to offer an explanation on how … More → The post Brief hijack makes Elsevier domains redirect to LAPSUS$ “Chapter II” page appeared first on Help Net Security .

Read the original article →

What happened

Cisco Talos researchers shared an open-source framework called CAIRN to help security practitioners classify and analyze malware that uses AI. [1]

CAIRN operates entirely on metadata extracted from files, without downloading or executing the malware. [2] [3]

Why it matters

CAIRN connects malware binaries through metadata attributes including submitter, import hash, domain, and AI provider. [4]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   To help security practitioners catch malware that leans on AI, researchers from Cisco Talos shared an open-source framework that they hope will be used to classify and analyze the threat.

2. [2]

   The tool, called CAIRN, works entirely from metadata pulled off files.

3. [3]

   No downloading the malware, no running it.

4. [4]

   CAIRN explorer connects malware binaries by metadata attributes like submitter, import hash, domain or AI provider (Source: Cisco Talos) How CAIRN hunts Researchers look for what Talos … More → The post Researchers uncover malware that uses AI to choose its next move appeared first on Help Net Security .

Read the original article →

What happened

Researcher Chaotic Eclipse released BigDiskBuster, a proof-of-concept exploit for a reported Windows Defender Update denial-of-service zero-day. [1] [2] [3]

The researcher claims BigDiskBuster can block Microsoft Defender from receiving platform and signature updates. [4] [5]

Why it matters

If the reported behavior works as claimed, blocking Defender platform and signature updates could interfere with its updating process; the source presents this as a proof-of-concept capability, not a confirmed universal effect. [4] [5] [6]

Known limitations

Chaotic Eclipse says the technique works on supported Windows versions, but describes the proof of concept as buggy and needing further development. [6]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   The researcher Chaotic Eclipse released BigDiskBuster, a PoC exploit for a Windows Defender Update DoS Zero-Day vulnerability.

2. [2]

   Security researcher Chaotic Eclipse , also known as INFINITE NIGHTMARE, MSNightmare and Nightmare-Eclipse, released a new zero-day exploit targeting Microsoft Defender.

3. [3]

   The researcher named the exploit BigDiskBuster, it triggers a Denial of Service Vulnerability in Windows Defender Update.

4. [4]

   The security researcher claims to have developed BigDiskBuster, a proof-of-concept that can block Microsoft Defender from receiving platform and signature updates.

5. [5]

   “This proof of concept is similar to UnDefend , it prevents windows defender from performing platform/signature updates.” wrote the expert.

6. [6]

   “Seems to work on all supported windows versions but PoC is a bit buggy and needs some rewritting but you get the idea.” Chaotic Eclipse says the technique works across supported Windows versions, although the current PoC is still buggy and needs further development.

Read the original article →

What happened

EvilTokens was a Microsoft device-code phishing kit used to compromise 12,000 email inboxes across more than 10,000 organizations and to bypass MFA for Microsoft 365 access. [1] [2]

Its AI chatbot analyzed victims’ inboxes to identify targets, trusted contacts to impersonate, and fraud strategies intended to increase criminals’ paydays. [3] [4]

Microsoft seized 50 websites, disabled more than 150 supporting domains, and notified affected customers to help remediate compromised accounts. [5] [6]

UK police arrested two alleged EvilTokens administrators; both were released on bail while the investigation continued. [7] [8]

Why it matters

Microsoft said the disruption does not eliminate the demonstrated model of an end-to-end AI-enabled cybercrime service. [9] [10]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   EvilTokens is a notorious Microsoft device-code phishing kit that emerged in February, and, within months of launching, had been used by criminals to compromise 12,000 email inboxes across more than 10,000 organizations worldwide.

2. [2]

   Like other similar phishing subscriptions, EvilTokens was sold as-a-service, and allowed buyers to bypass multi-factor authentication (MFA) and silently authenticate as the victim to the organization's Microsoft 365 applications.

3. [3]

   What made this one especially insidious, however, was its AI use.

4. [4]

   EvilTokens featured an AI chatbot that could analyze a victim’s inbox, and help criminals identify who to target, which trusted contacts to impersonate, and even which fraud strategies to use to maximize criminals’ paydays.

5. [5]

   Late last week, in a coordinated effort that spanned the US and UK, Microsoft seized 50 websites used to operate the service and disabled more than 150 additional domains tied to its supporting infrastructure.

6. [6]

   After receiving authorizations from the US District Court for the Eastern District of Virginia, Microsoft and Health-ISAC worked with Cloudflare, Coinbase, OpenAI, Railway, SpyCloud, The Shadowserver Foundation and TRM Labs, took down EvilTokens’ platform, and Microsoft notified affected customers, helping them remediate compromised accounts.

7. [7]

   Meanwhile, London’s Metropolitan Police Service on September 18 arrested two men, aged 32 and 38, who allegedly acted as the administrators of the EvilTokens website.

8. [8]

   Both men have been released on bail while the investigation continues.

9. [9]

   According to Steven Masada, associate general counsel and DCU GM, this is also DCU’s first action against an end-to-end AI-enabled cybercrime service.

10. [10]

    “The infrastructure supporting EvilTokens has been disrupted, but the model it demonstrated will not disappear with it,” he said in a blog shared with The Register ahead of publication.

Read the original article →

What happened

Microsoft and industry partners disrupted EvilTokens, a phishing-as-a-service platform linked to more than 12,000 compromised Microsoft customer inboxes across over 10,000 organizations; they seized 50 websites and disabled more than 175 supporting domains under a federal court order. [1] [2] [3]

EvilTokens used an AI-style chatbot to analyze victims’ inboxes, identify trusted relationships and payment authorizations, and support business-email compromise by stealing session tokens for inbox access and persistence. [4] [5]

Microsoft correlated at least 13 FBI Internet Crime Complaint Center complaints to EvilTokens-linked activity, representing approximately $1.7 million in reported losses, while characterizing that amount as conservative because incidents may be unreported or difficult to link. [9] [10] [11]

Coinbase traced about $1.1 million in EvilTokens revenue and identified more than 1,000 deposits from over 700 distinct addresses through June 2026; operators charged a $1,500 initiation fee and $500 recurring subscription. [12] [13] [14]

Why it matters

The service combined identity attacks, cloud-system tools, social engineering, and financial-fraud capabilities; researchers said it could map Microsoft Graph organizational structures and permissions, enabling lateral movement. [6] [7]

Microsoft reported that active tokens obtained through targeted phishing lures consistently bypassed multi-factor authentication, email gateways, and endpoint-security tools. [8]

Source-supported guidance

Experts advised organizations and employees to treat unsolicited device codes as a red flag, assume compromised accounts are fully cataloged within minutes, and independently verify requests to change payment information or redirect funds. [15]

Known limitations

Microsoft said the supporting infrastructure was disrupted, but warned that the broader criminal-service model demonstrated by EvilTokens will not disappear with the takedown. [16]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Microsoft, along with a group of industry partners, disrupted EvilTokens, a short-lived but highly consequential cybercrime platform that investigators linked to more than 12,000 compromised Microsoft customer email inboxes across more than 10,000 organizations globally, the company said Tuesday.

2. [2]

   Acting on federal court order Sept.

3. [3]

   15, Microsoft and partners seized 50 websites the phishing-as-a-service used for operations and disabled more than 175 domains linked to EvilTokens’ supporting infrastructure.

4. [4]

   The service was centered on an AI-style chatbot that cybercriminals used to analyze victims’ inboxes, identify trusted relationships, payment authorizations and other sensitive details that could facilitate fraud.

5. [5]

   It facilitated business-email compromise campaigns by stealing session tokens that allowed cybercriminals to sift through a victim’s inbox and maintain persistent access.

6. [6]

   EvilTokens significantly lowered the barrier to entry for cybercriminals by including specialized tools for identity attacks, cloud systems, social engineering and financial fraud in a single interface.

7. [7]

   The service allowed cybercriminals to map organizational structure and permissions in Microsoft Graph, which enabled lateral movement, researchers said.

8. [8]

   With active tokens gained through a collection of highly-targeted phishing lures, cybercriminals consistently bypassed multi-factor authentication, email gateways and endpoint security tools.

9. [9]

   “We cannot estimate the total fraud attributable to all EvilTokens activity.

10. [10]

    However, we were able to correlate at least 13 complaints filed with the FBI’s Internet Crime Complaint Center to EvilTokens-linked activity, representing approximately $1.7 million in reported losses,” a Microsoft spokesperson said.

11. [11]

    “Because many incidents go unreported and not all victims can be definitively linked to specific campaigns, we believe this is a conservative estimate.” Victims of EvilTokens were largely concentrated in the United States, Canada, the United Kingdom, Australia, India and France, according to Microsoft.

12. [12]

    Coinbase, which also aided the investigation into EvilTokens, said it traced about $1.1 million in revenue for EvilTokens from its paying customers.

13. [13]

    The virtual currency company’s threat researchers found more than 1,000 deposits to EvilTokens from more than 700 distinct addresses through June 2026.

14. [14]

    Operators sold access to the service through Telegram for a $1,500 initiation fee and a recurring $500 subscription.

15. [15]

    Experts advised organizations and employees to treat unsolicited device codes as a red flag, assume compromised accounts are fully cataloged in minutes, and independently verify requests to change payment information or redirect funds.

16. [16]

    “The infrastructure supporting EvilTokens has been disrupted, but the model it demonstrated will not disappear with it,” Masada warned.

Read the original article →

What happened

Researchers testing two inexpensive smart-glasses models found more than a dozen flaws across the glasses, companion app, and associated website. [1]

When powered on but not connected to the owner’s phone, the glasses could reportedly be claimed over Bluetooth without a password or meaningful pairing confirmation. [2]

After connecting, an attacker could reportedly control the glasses to capture photos or recordings, copy existing media, and intercept data sent between the glasses and phone. [3]

A Bluetooth-visible identifier, combined with an app-website weakness, could allegedly expose a user’s email address and date of birth. [4]

Voice or text submitted to the built-in AI, and images sent to it, were reportedly transmitted first to a server in Shenzhen and could be forwarded elsewhere; the tests did not establish subsequent data use. [5] [6]

Why it matters

Australia’s smart-device security standards apply to most consumer smart devices manufactured on or after March 4, 2026, but applicability depends partly on manufacture date and the rules have not yet been tested in a known enforcement action. [7] [8] [9]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   ABC Australia reports that researchers from NSB Cyber and Abstract Shield tested two inexpensive pairs, costing A$60 and A$110 (around US$42 and US$78), and found more than a dozen flaws across the smart glasses themselves, their app, and an associated website.

2. [2]

   The main problem they uncovered was insecure Bluetooth pairing: If the glasses were powered on and not connected to their owner’s phone, an attacker could connect first, with no password or meaningful pairing confirmation.

3. [3]

   After connecting, an attacker could reportedly control the glasses to capture photos or recordings, copy existing media, and intercept data moving between the glasses and the phone.

4. [4]

   The testing also found that a Bluetooth-visible device identifier could allegedly be used with a weakness in the app’s website to retrieve a user’s email address and date of birth.

5. [5]

   ABC reports that voice or text submitted to the built-in AI, along with images sent to it, was first transmitted to a server in Shenzhen and could be forwarded elsewhere, depending on the function.

6. [6]

   The tests did not establish how the data was subsequently used.

7. [7]

   Professor Kimberlee Weatherall, a technology regulation expert involved in the testing, said the failure to identify China in the privacy policy appeared to violate the Australian Privacy Principles , The timing of the testing is also relevant because Australia’s smart-device security standards apply to most consumer smart devices manufactured on or after March 4, 2026.

8. [8]

   Devices made before that date are outside the requirements.

9. [9]

   However, whether the new smart-device standards apply would depend partly on when the glasses were manufactured, and the rules have not yet been tested in a known enforcement action.

Read the original article →

What happened

President Donald Trump continued defending a hands-off approach to AI regulation after hacks carried out by U.S. commercial frontier models prompted calls for more oversight. [1] [2]

Trump emphasized U.S. AI dominance over geopolitical rivals, particularly China, and said he would not stifle AI growth; he has previously suggested that good leadership is the only regulation needed. [3] [4] [5] [6] [7]

Trump said the Department of Justice and other law-enforcement bodies could rein in companies that overstepped, but provided no specifics on enforcement, legal authority, or the boundary for intervention. [8] [9]

The administration established a voluntary testing regime for some commercial frontier models, largely at private industry’s discretion; the source says the order was delayed and altered and was followed by a separate non-public agreement on pre-release testing. [11] [12] [13]

Why it matters

Treasury Secretary Scott Bessent argued that private lawsuits and creator liability could incentivize stronger AI security, while opposing liability exemptions sought by laboratories. [10]

Critics argue that inadequate regulation or cybersecurity controls can make current AI systems difficult to fully control or monitor, while the administration argues regulation could harm U.S. innovation and competitiveness. [14] [15] [16]

AI adviser David Sacks defended limited restrictions by arguing that slowing U.S. development would not stop AI progress but could shift it to other countries, especially China. [23] [24] [25]

Known limitations

The source reports that experts view leading U.S. models as ahead of Chinese models, while cheaper Chinese lower- and middle-tier models can be more efficient and sometimes outperform more powerful models when given substantially more tokens. [17] [18] [19]

The source reports allegations that Chinese AI companies conduct widespread, systematic distillation of U.S. frontier models, with implicit encouragement from Beijing. [20] [21] [22]

Analyst Ronan Murphy said Washington and Beijing may share a view that AI adoption and its risks should be allowed to proceed, while noting there may be no formal agreement. [26] [27] [28] [29]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   President Donald Trump continued to defend his administration’s hands-off approach to AI regulation in the wake of hacks carried out by U.S.

2. [2]

   commercial frontier models that have rattled policymakers and industry veterans and spurred calls for more regulatory oversight.

3. [3]

   He also posited that nothing may matter more than future U.S.

4. [4]

   dominance of the technology over geopolitical rivals like China.

5. [5]

   “We are leading now over China, and everyone else, and I’m going to keep it that way!

6. [6]

   I’m not going to stifle Growth, of something that will be bigger than the Industrial Revolution, or the internet, itself.” Trump has previously suggested that good leadership is the only regulation the U.S.

7. [7]

   needs for artificial intelligence.

8. [8]

   He later claimed the Department of Justice was ready to “rein things in” if companies overstepped, but offered no specifics on enforcement, legal authority, or where he would draw that line.

9. [9]

   “We will be careful, and that’s why we have the Department of Justice, and other Law Enforcement bodies, that will rein things in if we have to, but I will only encourage AI or, SI (SUPER INTELLIGENCE)!” Trump concluded.

10. [10]

    Secretary of the Treasury Scott Bessent recently told Congress that private lawsuits could force AI companies to institute better security, saying it’s clear what the government “shouldn’t do on safety is to give these labs a liability exemption, which is what they are asking for.” “The best way to guarantee safety is that the creators are liable for what they build and generate,” Bessent said.

11. [11]

    “That’s not how we treat airlines or drug companies or food companies.” The Trump administration issued an executive order earlier this year that set up a voluntary testing regime for some commercial frontier models, largely at private industry’s discretion.

12. [12]

    That order was significantly delayed and altered by AI industry boosters to ensure that governmental review did not cause companies to postpone their release timelines for new models.

13. [13]

    That agreement did not last long before fast-moving events caused the administration to strike another, non-public agreement with frontier AI companies like OpenAI, Anthropic and others governing pre-release testing for models.

14. [14]

    Beyond existential fears, critics also argue that inadequate regulation or cybersecurity controls in current AI systems make them impossible to fully control or monitor.

15. [15]

    But the Trump administration has consistently argued that regulation will harm, not help, U.S.

16. [16]

    innovation and global competitiveness, and the threat of China frequently looms large in those discussions.

17. [17]

    Experts believe China’s AI models are behind U.S.

18. [18]

    models at the top of the market, where OpenAI and Anthropic have consistently pushed the frontier limits of model capabilities.

19. [19]

    But Chinese lower and “middle class” models are often cheaper, more efficient and can even outperform more powerful models because users can dedicate exponentially more tokens for their tasks.

20. [20]

    The U.S.

21. [21]

    government has accused Chinese AI companies of conducting widespread, “systematic” distillation of U.S.

22. [22]

    frontier models, with the implicit encouragement of Beijing.

23. [23]

    In defending the administration’s approach, David Sacks, co-chair of the President’s Council of Advisors on Science & Technology and a top adviser on AI issues, specifically cited the threat from China and other countries that he claimed would not be subject to similar restrictions.

24. [24]

    “We’re not the only country that has advanced AI labs, and as the president declared…we have to win this AI race,” Sacks told Politico in May, later adding “I think that’s the first thing to recognize is that if somehow we slow down or stop AI development, it doesn’t mean that AI progress is going to stop.

25. [25]

    It just means it’s going to happen in other countries and specifically China.” Some observers have alleged that despite their larger differences, top leaders in the U.S.

26. [26]

    and China may view AI similarly at the strategic level, specfically that increased adoption – and risks – of AI are inevitable.

27. [27]

    Ronan Murphy, director of the tech policy program at the Center for European Policy Analysis, posited that while there may not be a formal agreement between the two countries, “they share views both in Beijing and in Washington, particularly in the White House, of: you have to allow this to happen.” “Clearly there’s a call for regulation from many quarters of AI in the U.S.

28. [28]

    and elsewhere, but in the White House – and we heard David Sacks talking about it [recently] – It’s ‘let them cook,’ and the Chinese approach seems to be the same,” said Murphy in a press briefing.

29. [29]

    “So there might be consensus at that level, if nothing else.” The post Citing China, President Trump doubles down on hands-off approach to AI regulation appeared first on CyberScoop .

Read the original article →

What happened

ZCode was found packaging and git-encrypting entire user workspaces, including complete project histories, and uploading them to Alibaba Cloud. [1] [2]

The decryption key was held only by Z.ai’s server, so users could not access or delete the uploaded files. [3]

The reported behavior was linked to Repository Index: Repo Wiki generated pages in the cloud and triggered file uploads. [4] [5]

ZCode said the uploaded data had never been used to train its models. [6]

ZCode said external assessments found that previously uploaded data had been deleted and that Repo Wiki had been removed. [7] [8]

Why it matters

The researcher said users could not disable the behavior in settings and that ZCode’s privacy policy did not disclose it. [4]

Known limitations

The company said a full security assessment report would be released soon; the researcher criticized deletion of commit records and pre-patch upload code. [9] [10]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Chinese AI giant Z.ai has apologized after developers caught it pulling a Grok, packaging up and uploading user workspaces to cloud storage.

2. [2]

   In a case that’s highly reminiscent of the issues over which Elon Musk’s xAI was scrutinized in July, Z.ai’s code-generation harness wing, ZCode, was found packaging and git-encrypting entire user workspaces, including complete project histories, and shipping them off to Alibaba Cloud.

3. [3]

   Worse still, the private key used to decrypt the data was only held by the server under Z.ai’s control, meaning users could not access the files ZCode had uploaded, nor delete them.

4. [4]

   Ferstar, the researcher who first highlighted the issue, claimed there was no option for users to disable the behavior in their settings, and there was no disclosure of the practice in ZCode’s privacy policy.

5. [5]

   They said the core problem lay with the tool’s Repository Index functionality, which triggered the uploading of files after Repo Wiki generated pages in the cloud.

6. [6]

   ZCode released a statement on Monday apologizing for the “security issues” and confirming the data it uploaded had never been used to train its models.

7. [7]

   “We welcome developers to continue reviewing ZCode and reporting potential issues, and we will provide rewards based on the severity of the issues reported.” ZCode said it tasked the China Academy of Information and Communications Technology (CAICT) and Beijing security company NSFOCUS to probe its product following the implemented changes.

8. [8]

   The company claimed the two outside assessments concluded that all the previously uploaded data has now been deleted and said the Repo Wiki feature was removed.

9. [9]

   ZCode also open sourced the entire project on GitHub, “placing the code under community scrutiny and making ZCode more open and transparent.” “Once again, we sincerely apologize and welcome continued scrutiny from the community.

10. [10]

    The full security assessment report will be released soon.” Ferstar confirmed the open sourced code showed no signs of the Repo Wiki still being implemented, but criticized the company for wiping commit records and the source code ZCode used to upload files pre-patch.

Read the original article →

What happened

A proof-of-concept tool named BigDiskBuster was published on GitHub on September 19; it fills available disk space and stops Microsoft Defender from installing platform and signature updates. [1]

The tool had no patch, CVE, or Microsoft advisory at the time described by the source. [2]

The source identifies Abdelhamid Naceri as the tool’s author and a former Microsoft security researcher. [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A zero-day proof-of-concept tool that stops Microsoft Defender from installing platform and signature updates by filling all available disk space was published on GitHub on September 19.

2. [2]

   The tool, called BigDiskBuster, has no patch, no CVE, and no Microsoft advisory.

3. [3]

   Its author, Abdelhamid Naceri, is a former Microsoft security researcher whose earlier Defender exploits were used in

Read the original article →

What happened

Security researcher Abdelhamid Naceri, known as NightmareEclipse, released the BigDiskBuster proof of concept to prevent Microsoft Defender Antivirus from installing platform and security intelligence updates. [1] [2]

The proof of concept waits for a Defender update, consumes available drive space with hidden temporary files so the update cannot finish, then releases the space after detecting failure. [3] [4] [5] [6]

BigDiskBuster also opens MRT.exe while restricting other processes' access to the executable for as long as its handle remains open. [7]

Why it matters

The reported effect is to leave Defender on its current platform and security-intelligence versions while the tool interferes with updates; the antivirus may continue running but could be less able to identify newly detected malware without current definitions. [8] [9] [10]

Known limitations

The researcher claims BigDiskBuster works on all supported Windows versions, but the current proof of concept is described as buggy and that compatibility claim has not been independently verified. [11]

A screenshot shows Defender reporting update error 0x80070643, but that generic installation error does not by itself establish that BigDiskBuster caused it. [12] [13]

There is currently no indication that BigDiskBuster has been used in real-world attacks. [14]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Serial Microsoft zero-day leaker NightmareEclipse has found another way to mess with Windows Defender, this time by stopping the antivirus from updating itself.

2. [2]

   The security researcher, also known as Abdelhamid Naceri, released a proof-of-concept dubbed “BigDiskBuster” that is designed to prevent Microsoft Defender Antivirus from installing platform and security intelligence updates.

3. [3]

   The trick doesn't disable Defender.

4. [4]

   Instead, the PoC waits for an update to start, then tries to fill up the drive so there isn't enough space for it to finish.

5. [5]

   The code does this by creating hidden temporary files sized to consume the drive's free space, spinning up additional threads as needed to claim more.

6. [6]

   Once it detects that the Defender update has failed, it closes the files and returns the space.

7. [7]

   BigDiskBuster also opens Microsoft's Malicious Software Removal Tool executable, MRT.exe, in a way that restricts other processes' access to the file while the handle remains open.

8. [8]

   The result, according to NightmareEclipse, is that Defender stays stuck on its current platform and security intelligence versions as long as the tool keeps interfering with updates.

9. [9]

   Leaving Defender stuck on old security intelligence is obviously less than ideal.

10. [10]

    The antivirus may still be running, but preventing it from receiving Microsoft's latest threat definitions could leave it less able to identify newly detected malware.

11. [11]

    The researcher describes BigDiskBuster as similar to their earlier “UnDefend” tool and claims it works on all supported versions of Windows, although they admit the current PoC is “a bit buggy and needs some rewritting [sic].” That compatibility claim has not been independently verified.

12. [12]

    A screenshot published alongside the PoC shows Windows Security reporting that a protection definition update failed with error 0x80070643.

13. [13]

    That's a generic installation error, however, and isn't evidence on its own that BigDiskBuster is at work.

14. [14]

    There’s currently no indication that BigDiskBuster has been used in real-world attacks, and NightmareEclipse's claim that it works across all supported Windows versions remains unverified.

Read the original article →

What happened

Bifrost, an open-source AI gateway routing requests to more than 20 LLM providers, has a critical vulnerability allowing an unauthenticated attacker to run arbitrary commands on the gateway server with one HTTP request. [1]

The vulnerability is tracked as CVE-2026-90898 and has a CVSS score of 9.8. [2]

The flaw affects Bifrost HTTP transport versions before 2.1.0; the supplied evidence truncates the statement describing the management-authentication condition. [2]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A critical vulnerability in Bifrost, an open-source AI gateway that routes requests to more than 20 LLM providers, allows an unauthenticated attacker to run arbitrary commands on the gateway server with a single HTTP request.

2. [2]

   The flaw, tracked as CVE-2026-90898 (CVSS score: 9.8), affects all versions of the Bifrost HTTP transport before 2.1.0 when management authentication is

Read the original article →

What happened

ShinyHunters claims it exploited an Oracle PeopleSoft zero-day on the FBI jobs webpage, enabling remote code execution on servers; Oracle had not confirmed awareness of the alleged vulnerability. [1] [2]

The group claims it defaced the FBI website and moved laterally from the compromised site to FBI-managed servers in AWS GovCloud. [3] [4]

ShinyHunters claims it downloaded approximately 2 TB to 3 TB of data belonging to current, former, and prospective FBI employees, including data from human resources, MedLink, and Criminal Justice Information Services. [4] [5] [6]

Why it matters

The group says the activity was not financially motivated and instead sought retractions or corrections to FBI statements about its operations. [7] [8] [9] [10]

Known limitations

The reported intrusion, data theft, PeopleSoft zero-day, and scope of affected services were claims by ShinyHunters; the FBI, Oracle, and AWS had not immediately responded to the publication’s inquiries. [1] [2] [4] [6] [8]

The FBI previously characterized ShinyHunters as using harassment, threats, swatting, and potentially false claims about victims’ sensitive information; ShinyHunters denied those characterizations. [11]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   According to a ShinyHunters spokesperson, the extortion group exploited an Oracle PeopleSoft zero-day vulnerability on the FBI jobs webpage, which it says allowed remote code execution (RCE) on the servers.

2. [2]

   Neither Oracle nor AWS immediately responded to our inquiries, including whether Oracle is aware of a PeopleSoft preauth RCE zero-day, and whether AWS has any insight into the alleged data theft.

3. [3]

   The group then defaced the website, replacing it with a “This site has been seized by ShinyHunters” banner and image shared with The Register.

4. [4]

   At press time, the site says it is “currently down for maintenance but will be back up soon!” ShinyHunters also claims it moved laterally from the compromised site onto the FBI’s managed servers on AWS GovCloud, and downloaded about 2 TB to 3 TB of data belonging to current, former, and prospective FBI employees.

5. [5]

   “We hold data on all FBI employees and applicants,” the spokesperson told us.

6. [6]

   ShinyHunters claims the compromised FBI services include human resources, MedLink, and Criminal Justice Information Services.

7. [7]

   “This is NOT financially motivated,” a Shiny spokesperson told The Register.

8. [8]

   “We want the FBI to correct or retract their statements they made, which included substantial false allegations.” The FBI did not immediately respond to The Register’s request for comment.

9. [9]

   Unlike most of the group’s smash-and-grab operations that involve a multimillion-dollar ransom demand to not leak the stolen files, ShinyHunters said it isn't seeking an extortion payment from the FBI.

10. [10]

    Instead, it wants the federal cops to retract statements made about ShinyHunters in a May 15 bulletin, shortly after the gang broke into ed-tech giant Instructure's Canvas platform and claimed to have stolen data tied to hundreds of millions of students, teachers, and staff.

11. [11]

    The FBI said ShinyHunters uses “harassment strategies, sending threatening text messages and phone calls to victims and their family members, and in some cases, swatting.” The security alert also said that extortionists “may falsely claim to have sensitive or compromising information, including embarrassing photographs or videos of victims, which frequently do not exist.” Shiny claims none of this is true.

Read the original article →

What happened

Researchers disclosed a malicious npm package, tw-pkgprobe-7731, that masquerades as a security tool for developers integrating Twilio applications while attempting to harvest sensitive data. [1]

tw-pkgprobe-7731 was first uploaded to the npm registry in mid-August 2026 by the account twdepprobe7731. [2]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Cybersecurity researchers have disclosed details of a malicious npm package named "tw-pkgprobe-7731" that masquerades as a security tool targeting developers integrating Twilio into their applications, while stealthily attempting to harvest sensitive data.

2. [2]

   The package, named "tw-pkgprobe-7731," was first uploaded to the npm registry in mid-August 2026 by an npm account named "twdepprobe7731."

Read the original article →

What happened

WordPress fixed a critical core flaw that allows an unauthenticated attacker to make a site load a PHP file from outside its theme folders. [1]

The fix shipped on September 22 in WordPress 7.1.2 and covers every still-supported branch back to 4.7. [3]

Why it matters

On some servers, exploitation can allow the attacker to run their own code. [2]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   WordPress has fixed a critical flaw in its core software that lets an attacker with no account make a site load a PHP file from outside its theme folders.

2. [2]

   On some servers, that can go further, allowing the attacker to run their own code.

3. [3]

   The fix shipped on September 22 in WordPress 7.1.2, with fixes for every branch the project still supports, back to 4.7, and WordPress is telling site owners

Read the original article →

What happened

Check Point said attackers exploited a previously unknown flaw in its Security Management Server in a handful of targeted attacks on July 23. [1]

CVE-2026-93616 allows an attacker who can access the server’s web service to run scripts without logging in. [2]

Check Point released a fix on September 22 for the affected server, which controls Check Point firewall policies. [3]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Attackers exploited a previously unknown flaw in Check Point's Security Management Server in a handful of targeted attacks on July 23, the company said.

2. [2]

   The flaw, CVE-2026-93616, allows an attacker who can access the server's web service to run scripts on it without logging in.

3. [3]

   Check Point released a fix on September 22 for the server that controls firewall policies for the Check Point

Read the original article →

What happened

Volexity reported that the China-aligned group UTA0565 exploited a three-vulnerability zero-day chain affecting Chrome and Microsoft before the defects were disclosed or patched, during September 3–4. [1] [2] [3]

The chain included CVE-2026-85046 and CVE-2026-87491, remote-code-execution defects in Chromium-based browser JavaScript engines, and CVE-2026-85880, a Windows Advanced Local Procedure Call privilege-escalation zero-day. [4] [5]

Why it matters

Volexity said UTA0565 used multiple fake websites, phishing emails targeting Asian government entities, and spoofed domains impersonating the Center for American Progress and China Digital Times. [6] [7] [8]

Volexity reported that UTA0565 used the previously undocumented CLEANGULP malware family and found likely related domains targeting media, halal restaurant search, and corporate training organizations. [9]

Volexity wrote that the exploit kit appeared to be adopted across multiple Chinese threat actors and was likely shared, customized, and weaponized; it also cautioned that observations from only two organizations may understate the activity’s scope and impact. [4] [10] [11]

Known limitations

Proofpoint reported that only a limited group of organizations were exposed to all three vulnerabilities within a short window and had previously attributed related attacks to APT31, UNK_LateNight, UNK_DoubleCheck, and UNK_QuietRacket. [12] [13]

Volexity characterized UTA0565’s zero-day use as technically and operationally improved compared with other campaigns it had observed, while noting that legitimate website content remained effective decoy material for reducing user suspicion. [14] [15]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Volexity researchers spotted another state-aligned Chinese threat group exploiting a triple-link chain of zero-day vulnerabilities across multiple campaigns, the company said in a blog post Monday.

2. [2]

   The threat group it tracks as UTA0565 exploited the vulnerabilities in Chrome and Microsoft between Sept.

3. [3]

   3 and 4 before the defects were disclosed or patched, researchers said.

4. [4]

   “The activity reported so far reflects only two organizations’ observations; the full scope and impact are likely far broader.” The vulnerabilities include: CVE-2026-85046 and CVE-2026-87491 , remote-code execution defects in the JavaScript engine for Chromium-based browsers; and CVE-2026-85880 , a privilege-escalation zero-day that Microsoft disclosed Sept.

5. [5]

   8 in Windows Advanced Local Procedure Call.

6. [6]

   Yet, Volexity noted UTA0565’s campaigns differed from those attacks by using multiple fake websites to deceive victims.

7. [7]

   Volexity shared phishing emails UTA0565 sent to Asian government entities urging them to publicly support imprisoned Hong Kong activist Chow Hang-tung.

8. [8]

   The group spoofed domains impersonating the Center for American Progress and China Digital Times in other phishing emails.

9. [9]

   Volexity said UTA0565 used a payload from a previously undocumented malware family it tracks as “CLEANGULP.” Researchers also found several domains likely used by UTA0565 in similar campaigns targeting media organizations, halal restaurant search websites and corporate training organizations.

10. [10]

    While UTA0565 showcased a variance in tactics, it used the same components researchers observed in previous instances of the exploit kit across multiple Chinese threat groups.

11. [11]

    “This seemingly widespread adoption across multiple threat actors suggests a coordinated effort within the Chinese computer network exploitation community, where the core kit was likely shared, customized, and weaponized by multiple groups,” Volexity wrote in the blog post.

12. [12]

    Proofpoint, which previously observed multiple state-aligned threat groups chaining the vulnerabilities together in attacks since last August, said a limited group of organizations were exposed to all three vulnerabilities in a short window.

13. [13]

    Proofpoint previously attributed attacks involving the zero-days to APT31, UNK_LateNight, UNK_DoubleCheck and UNK_QuietRacket.

14. [14]

    “UTA0565’s use of the zero-day vulnerabilities shows technical and operational improvements over other campaigns observed by Volexity, both in the mechanics of the exploitation and the presentation to end users,” researchers wrote.

15. [15]

    “Using real content from legitimate websites as decoy material continues to be an effective way to reduce user suspicion.” The post Volexity spots another China-aligned threat group exploiting Chrome and Microsoft defects appeared first on CyberScoop .

Read the original article →

What happened

Microsoft said it led an industry-wide disruption of EvilTokens, a subscription-based scam platform that used an AI chatbot and compromised 12,000 Microsoft accounts over a few months. [1]

Introduced through Telegram in February, EvilTokens charged an initial $1,500 fee and $500 per month thereafter. [2]

The platform streamlined much of the process for compromising email accounts at scale, including inbox analysis, target selection and drafting follow-up fraud emails. [3] [4]

Why it matters

Microsoft said the chatbot analyzed victims’ inboxes to identify trusted relationships, payment authorizations and sensitive responsibilities where fraud was most likely to succeed. [5]

Microsoft said the platform could recommend fraud strategies and draft messages impersonating trusted contacts to induce victims to act. [6]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Microsoft said Tuesday that it led an industry-wide disruption of a subscription-based scam platform that used an AI chatbot to compromise 12,000 Microsoft accounts over a few-month span.

2. [2]

   Named EvilTokens, the platform was introduced over a Telegram channel in February and charged an initial $1,500 fee and a recurring $500 charge each month after that.

3. [3]

   EvilTokens provided a single service for streamlining most steps required to compromise email accounts in large numbers.

4. [4]

   From there, the platform helped customers analyze inboxes, select targets that would provide the biggest potential payouts, and draft follow-up emails that provided realistic ruses for tricking company employees into transferring funds to attacker-controlled accounts.

5. [5]

   Minutes, not days “While EvilTokens helped cybercriminals access email accounts, at the center of the service was an AI-style chatbot that could analyze a victim’s inbox and help criminals identify trusted relationships, payment authorizations, and sensitive responsibilities, as well as other circumstances where fraud was most likely to succeed,” Microsoft said .

6. [6]

   “The platform could even recommend fraud strategies, including drafting messages that impersonated trusted contacts to help criminals trick victims into taking action.” Read full article Comments

Read the original article →

What happened

Rep. Josh Gottheimer introduced the AI Cyber Defense Act to establish a CISA test program giving critical-infrastructure operators access to frontier AI models and technical assistance for protecting, detecting, testing, and remediating cybersecurity vulnerabilities. [1] [2] [3]

The bill would authorize $100 million for the pilot from 2027 through 2031, but the funding would depend on appropriators providing the money. [4]

Critical-infrastructure owners and operators could apply, with priority directed to nonprofit, publicly owned, rural, and small-sized organizations. [5]

Why it matters

The proposal was inspired by cyberattacks on water facilities, and Gottheimer said communities may lack resources to pay for AI-based vulnerability analysis and patching. [2] [5] [6]

The proposal is different from a recently announced Texas test program; the article reports criticism that the Texas program relied on voluntary private-sector participation and lacked a significant budget. [7] [8]

Known limitations

The article reports that federal funding for critical infrastructure has an uncertain future and that CISA funding was significantly cut during the administration’s second term. [6] [9]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A House Democrat tapped to lead his party’s efforts on artificial intelligence has introduced legislation that would establish a test program within the Cybersecurity and Infrastructure Security Agency to give critical infrastructure operators free access to frontier AI models to protect their systems.

2. [2]

   Josh Gottheimer, D-N.J., introduced the AI Cyber Defense Act Monday, inspired by the series of cyberattacks on water facilities in recent months.

3. [3]

   The bill directs the Department of Homeland Security, through CISA, to create a program “through which owners and operators of critical infrastructure that participate in the Program are able to securely utilize artificial intelligence procured through the Secretary and technical assistance provided by the Secretary to protect against, detect, test for, and remediate vulnerabilities in the cybersecurity of such critical infrastructure.” AI-tinged, water-focused cybersecurity pilot programs are all the rage lately.

4. [4]

   Gottheimer’s bill would authorize $100 million for the pilot program from 2027 to 2031 before it ends, although appropriators would have to follow through on providing the actual dollars.

5. [5]

   “It’s expensive to bring the AI in to analyze your system and find those vulnerabilities.” Critical infrastructure owners and operators could apply for the pilot program, which the bill directs to give priority to nonprofit, publicly owned, rural and small-sized organizations.

6. [6]

   “Right now federal funding for critical infrastructure has an uncertain future and many of our local communities just don’t have the resources they need to pay for AI tokens to do the patching they need,” Gottheimer said when he introduced the bill.

7. [7]

   The introduction of Gottheimer’s legislation is adjacent to, but different from, a test program that the Office of the National Cyber Director recently announced in Texas.

8. [8]

   One criticism of that program is that private sector companies offered their cyber and AI services through it on a purely voluntary basis, with no significant budget to bolster the pilot.

9. [9]

   The Trump administration has significantly cut CISA funding in its second term.

Read the original article →

What happened

Cisco Talos describes CLOSEDQUORUM as a Windows malware implant that queries up to four LLM providers—Gemini, DeepSeek, Qwen, and Mistral—to select predefined post-compromise actions, including credential and cryptocurrency-wallet theft. [1] [2]

After deployment, CLOSEDQUORUM reportedly operates without continued human commands; Talos calls it, to its knowledge, the first publicly documented Windows implant using this approach for C2. [3]

The LLM-selected modules are constrained to executable predefined actions: Steal, Inject, and Persist. Steal can dump LSASS memory, browser passwords, and cryptocurrency-wallet data; Inject uses shellcode with process hollowing or Early Bird injection; Persist establishes persistence. [4] [5] [6] [7]

Why it matters

Talos analyst Ryan Fetterman said transferring an attack phase from a human operator to AI systems can remove a human bottleneck, allowing activity to continue when an operator is not watching; this was presented as a potential speed and scale advantage. [8] [9] [10]

Known limitations

Talos has not observed CLOSEDQUORUM deployed in the wild, although binary artifacts reportedly link its developer to carding-related criminal-forum postings dating to 2025. [11]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A new Windows malware called CLOSEDQUORUM can query up to four LLM providers - Google Gemini, DeepSeek, Qwen, and Mistral - to autonomously select from predefined post-compromise actions, including stealing users’ credentials and cryptocurrency wallets.

2. [2]

   After deployment, the Go-based malware delegates its next action to a quorum of LLMs that vote on what it should do next.

3. [3]

   Once deployed, the malware does not require continued commands from a human operator, according to Cisco Talos, which describes it as, to its knowledge, the first publicly documented Windows implant to use this approach for command-and-control (C2).

4. [4]

   It does not go offline when the attacker sleeps.” The models’ decisions are limited to the pre-defined actions, and they must choose “ONLY executable decisions,” according to a system prompt that Talos’ researchers extracted from the binary.

5. [5]

   It tells each model: “You are an advanced malware strategist.” And then the models choose what the malware should do from these capability modules: Steal, which simultaneously runs commands to dump LSASS memory for Windows credentials, steal saved browser passwords across Google Chrome, Microsoft Edge, and Mozilla Firefox, and extract cryptocurrency wallet data including MetaMask, Exodus, and Ethereum.

6. [6]

   Inject generates shellcode and then uses process hollowing or Early Bird injection to execute malicious code.

7. [7]

   Persist establishes persistence on the infected device.

8. [8]

   “Four models are queried in sequence, their independent verdicts tallied, and the binary acts, based on their judgment.” This type of “effort displacement,” which transfers a phase of the attack from a human operator to AI systems, can compound the speed and scale advantages of an intrusion by removing the human bottleneck, Fetterman added.

9. [9]

   “Human operators are bound by attention, working hours, and cognitive load,” he wrote in the Tuesday blog.

10. [10]

    “An AI system capable of executing a phase of the attack chain can continue when the operator is no longer watching.

11. [11]

    While the threat hunters haven’t observed any in-the-wild deployment of CLOSEDQUORUM, they said that artifacts from the binary link the malware’s developer to postings that date back to 2025 on criminal forums related to carding.

Read the original article →

What happened

The FBI said it is investigating unauthorized activity affecting FBIjobs.gov after ShinyHunters claimed responsibility; the breach temporarily defaced the FBI jobs site. [1] [2] [3]

ShinyHunters claimed it stole sensitive data on almost all FBI agents and people who applied for FBI jobs, but the FBI’s statement did not confirm that claim. [3] [4]

The apply.fbijobs.gov site and Special Agent Application Portal were reported unavailable following the attack. [5]

Why it matters

Flashpoint analysts said ShinyHunters has previously overstated the criticality of accessed data but has established itself as a legitimate threat. [6]

The group did not directly threaten to release the allegedly stolen data, instead setting a one-week deadline for action; the article says it appeared to seek changes to the FBI’s May public service announcement rather than a payoff. [7] [8]

The attack followed an FBI public service announcement that ShinyHunters said contained false allegations about the group; the FBI issued that PSA after a May attack on Instructure. [9] [10]

The article describes ShinyHunters as having previously targeted cloud platforms, healthcare organizations, universities, technology companies, retailers and education providers. [11]

Known limitations

The supplied evidence does not establish whether ShinyHunters obtained the claimed FBI agent and applicant data or whether the data will be released. [4] [8]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   The FBI is investigating an attack on its own systems after ShinyHunters claimed responsibility for the incident, putting the prolific cybercrime group in the most direct conflict yet with agents responsible for investigating data extortion attacks.

2. [2]

   The Monday breach, first reported by 404 Media , allowed ShinyHunters to temporarily deface the FBI jobs site.

3. [3]

   “The FBI is aware of claims regarding unauthorized activity affecting FBIjobs.gov and is currently investigating,” a spokesperson for the agency said in a statement.

4. [4]

   The group claimed it stole “very sensitive data on almost all FBI agents and individuals who filed an application with the FBI for a job,” in a lengthy post on its data-leak site.

5. [5]

   An alert on the FBI jobs site notes that apply.fbijobs.gov and the Special Agent Application Portal are currently unavailable.

6. [6]

   “While ShinyHunters has in the past been hyperbolic about the criticality of the data they’ve accessed, the group has established itself as a legitimate threat,” Flashpoint analysts told CyberScoop.

7. [7]

   ShinyHunters doesn’t appear to be seeking a payoff in this case, but rather a bid to coerce the FBI into amending or removing the May PSA.

8. [8]

   The group didn’t make any direct threat in the data-leak site post to release the stolen data, but it set a deadline of one week for action.

9. [9]

   ShinyHunters claims it targeted the FBI in response to a public service announcement it says contains false allegations about the group.

10. [10]

    The FBI issued the PSA following ShinyHunters’ May attack on Instructure, the company behind Canvas, a widely used central hub for K-12 and university coursework, exams and communication.

11. [11]

    The attack marks a sobering escalation by ShinyHunters, a notorious group that previously targeted major cloud platforms, healthcare organizations, universities , technology companies, retailers and education service providers.

Read the original article →

What happened

A remotely triggerable null-pointer dereference in NetBSD’s ipfilter can be triggered from outside the machine and may cause the kernel—and therefore the system—to go down. [1] [2] [3]

The NetBSD Project shipped a fix for the ipfilter kernel flaw in NetBSD 10.2 on September 15. [4]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   A NetBSD box at the edge of a network, filtering traffic with ipfilter, has been carrying a kernel flaw that someone outside the machine can set off.

2. [2]

   The bug is a remotely triggerable null pointer dereference in ipfilter, meaning the kernel tries to read memory through a pointer that leads nowhere.

3. [3]

   In kernel code, that usually ends with the whole system going down.

4. [4]

   The NetBSD Project shipped the fix on September 15 in NetBSD 10.2, … More → The post NetBSD 10.2 security fixes close a remote kernel bug in ipfilter appeared first on Help Net Security .

Read the original article →

What happened

Rabbit released OS3, an agentic operating system for the r1 handheld that runs in Rabbit’s cloud and operates a user’s computers through a locally installed agent. [1] [2]

Users state goals in chat; OS3 selects a connected device and can work with desktop software, local files, web pages, or written and executed code to complete the task. [3]

One Rabbit account can connect up to five devices. [4]

Cited source evidence

These excerpts support the numbered claims above. They come from one source article and do not establish independent corroboration.

1. [1]

   Rabbit, the Santa Monica company that makes the r1 handheld, released OS3, an agentic operating system.

2. [2]

   It runs in Rabbit’s cloud and operates a user’s computers through a local agent that installs with one command.

3. [3]

   The user states a goal in a chat, and OS3 picks the device, then works desktop software, local files and web pages, or writes and runs code, to complete the task.

4. [4]

   One account can connect up to five devices, including … More → The post Rabbit’s new OS lives in the cloud and borrows your laptop to get things done appeared first on Help Net Security .

Read the original article →