November 5, 2026

What happened

Okta’s 2026 Customer Identity Trends Report says customer control, account security and ease of use influence adoption of AI agents that make purchases, book services and access personal information. [1]

Consumers’ willingness to delegate depends on an agent’s capabilities, the consequences of mistakes, and whether users can approve or reverse its actions. [2]

Why it matters

Businesses introducing these services must meet new requirements for verifying agents and managing their permissions. [3]

Known limitations

The available excerpt states that only 37% of consumers are referenced, but the sentence is truncated and does not provide the measure or its context. [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]

   Customer control, account security and ease of use are influencing the adoption of AI agents that make purchases, book services and access personal information, according to Okta’s 2026 Customer Identity Trends Report.

2. [2]

   Consumers’ willingness to delegate depends on what an agent can do, the consequences of a mistake and their ability to approve or reverse its actions.

3. [3]

   Businesses introducing these services must meet new requirements for verifying agents and managing their permissions.

4. [4]

   Only 37% of … More → The post Most consumers aren’t ready to let AI agents shop unsupervised appeared first on Help Net Security .

Read the original article →

What happened

DNSFilter’s report states that 74.5% of respondents had larger cybersecurity budgets than 12 months earlier. [1] [2]

Among respondents’ next incremental security spending priorities, AI security and governance ranked first at 40.3%, followed by cloud security at 25.5%. [1] [3]

Why it matters

The article identifies employee permissions granted to AI tools as a stated reason for prioritizing AI security and governance, but the supplied excerpt truncates the supporting explanation. [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]

   Companies are increasing their cybersecurity budgets, with AI security and governance the top priority for additional spending, followed by cloud security, according to DNSFilter’s The Cost of Shadow AI report.

2. [2]

   Some 74.5% of respondents reported a larger cybersecurity budget than 12 months earlier.

3. [3]

   Asked where their next incremental security dollar would go, 40.3% named AI security and governance, followed by cloud security at 25.5%.

4. [4]

   One reason to prioritize it is that employees grant AI tools … More → The post Companies are spending more on AI governance and getting less control appeared first on Help Net Security .

Read the original article →

What happened

According to Paubox, nearly nine in ten healthcare leaders say security or compliance staff always or usually review AI-written code touching patient data before deployment. [1]

Paubox also reports that seven in ten healthcare leaders have found an AI-built tool handling patient data without IT or security awareness. [1]

Why it matters

The reported tools include scripts, automations, and integrations created by staff with AI coding assistants. [2]

The source states that code review covers only what someone submits, while the article identifies undiscovered AI tools as a governance problem. [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]

   Nearly nine in ten healthcare leaders say security or compliance staff always or usually review AI-written code that touches patient data before it goes live, and seven in ten have still found an AI-built tool handling that data without IT or security knowing, according to Paubox.

2. [2]

   Source: Paubox These are the scripts, automations, and integrations staff put together with AI coding assistants.

3. [3]

   Here’s the catch: a code review only covers what someone submits, and a … More → The post Healthcare can’t govern the AI tools it hasn’t found yet appeared first on Help Net Security .

Read the original article →

What happened

Ryan James Roach was extradited to the United States to face federal charges related to an alleged hacking operation targeting universities and other schools. [1] [2]

The indictment alleges that, from June 2016 through December 2019, Roach and Mathiew James Stubbings hacked computers, stole login details, and used victims’ machines to mine cryptocurrency. [3]

According to the Justice Department, the alleged operation used virtual private servers registered under false names to transmit malware and mining software and store stolen credentials. [4] [5]

Prosecutors allege the defendants accessed computers without authorization, deployed malware, stole and exfiltrated credentials, and used compromised systems’ processing power for their financial benefit. [10]

The indictment alleges network reconnaissance, use of servers and compromised machines to conceal activity, and movement through the network after initial access. [11] [12]

The charges include conspiracy involving protected computers, obtaining information from a protected computer, and intentional damage to a protected computer; each charge carries up to 10 years in prison and a $250,000 fine. [13] [14]

Why it matters

The alleged victims included educational institutions in New York, Tennessee, North Carolina, and Massachusetts, including a higher-education institution in western New York. [6] [7] [8] [9]

Source-supported guidance

The source advises security teams to patch quickly, monitor for unusual processing loads, and treat credential theft as a serious incident. [15] [16] [17]

Known limitations

The allegations remain unproven: the Justice Department states that charged defendants are presumed innocent unless and until proven guilty. [18]

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 Canadian man, Ryan James Roach (35), has been extradicted to the United States to face charges tied to a hacking operation that targeted universities and other schools.

2. [2]

   Roach was extradited on October 28 and now faces federal charges alongside Mathiew James Stubbings, 41, of Ontario.

3. [3]

   The indictment alleges that, from June 2016 to December 2019, the two men hacked computers, stole login details, and used victims’ machines to mine cryptocurrency for themselves.

4. [4]

   “They used virtual private servers in multiple countries, which were registered under false names.

5. [5]

   These servers transmitted malware and mining software to victim computers and stored stolen login credentials.” DoJ states .

6. [6]

   “Victims of the scheme included educational institutions in New York, Tennessee, North Carolina, and Massachusetts.

7. [7]

   Some of the computers damaged by the defendants were located at an institution of higher learning in the Western District of New York.” The targets were not random home PCs.

8. [8]

   Prosecutors say the pair went after U.S.

9. [9]

   Some affected computers belonged to a higher-education institution in western New York.

10. [10]

    “To carry out the scheme, Roach and Stubbings surreptitiously accessed victim computers without authorization, deployed malware to gain control, stole victim login credentials, exfiltrated the data to servers that they controlled, and forcing victim computers to use their processing power to mine cryptocurrency for the financial benefit of the defendants.” DoJ continues.

11. [11]

    The indictment says they looked for weaknesses in victims’ networks, used servers and compromised machines from one victim to hide their tracks, and then moved through the network once inside.

12. [12]

    The duo used sophisticated techniques and tools to conduct reconnaissance and expand their access within the network.

13. [13]

    The charges include conspiracy to obtain information from and damage protected computers, obtaining information from a protected computer, and intentional damage to a protected computer.

14. [14]

    Each carries up to 10 years in prison and a $250,000 fine.

15. [15]

    For security teams, the useful part is not the alleged crypto-mining.

16. [16]

    It is the reminder that universities remain attractive targets because they hold powerful computing resources, broad networks, and plenty of people with access.

17. [17]

    Patch quickly, watch for unusual processing load, and treat credential theft as a serious incident rather than a nuisance.

18. [18]

    “The fact that a defendant has been charged with a crime is merely an accusation and the defendant is presumed innocent until and unless proven guilty.” DoJ concludes.

Read the original article →

What happened

The interview describes AI agents as goal-directed systems that choose their own steps, unlike scripted bots, and says this makes them harder to detect. [1] [2]

Zohaib Ahmed, CEO of Resemble AI, discusses agents arriving as declared, undeclared, or disguised traffic and says binary bot detection misses the latter two groups. [3]

The interview identifies click timing and groups of sessions pursuing one goal as signals that may reveal agents. [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]

   Bots follow a script.

2. [2]

   AI agents work toward a goal and choose their own steps, which makes them harder to catch.

3. [3]

   In this Help Net Security interview, Zohaib Ahmed, CEO of Resemble AI explains how agents arrive as declared, undeclared or disguised traffic, and why binary bot detection misses the last two groups.

4. [4]

   Ahmed walks through the signals that give agents away, like click timing or a batch of sessions all chasing one goal.

Read the original article →

What happened

Binarly found that Zyxel GS1900 switches receiving an earlier fix for an actively exploited flaw remained open to the exploit used by attackers. [1]

Anyone able to reach a switch’s web management interface can reportedly run commands on it without logging in. [2]

Zyxel released new firmware for 10 GS1900 models that closes the hole. [3]

Known limitations

The supplied evidence truncates the identification of the earlier fix after “2.90,” so its full version and applicability are unknown. [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]

   Zyxel GS1900 network switches that received the vendor's earlier fix for an actively exploited flaw are still open to the exploit attackers used, researchers at Binarly found.

2. [2]

   The exploit lets anyone who can reach a switch's web management interface run commands on it without logging in.

3. [3]

   Zyxel has released new firmware for 10 GS1900 models that closes the hole.

4. [4]

   The earlier fix is the 2.90

Read the original article →

What happened

Apple will begin checking iCloud content shared through “Anyone with the link” in iOS, iPadOS, macOS, and visionOS 27.3 or later for harmful material. [1]

For iCloud content sharing, users can choose “Share a Link” or “Send a Copy”; “Share a Link” supports restricting access to invited or approved people or allowing anyone with the link to access the content. [2]

Known limitations

The supplied evidence does not specify what material will be checked, how checks operate, or what account penalties may apply. [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]

   Apple will begin checking iCloud content shared using the “Anyone with the link” option for harmful material in iOS, iPadOS, macOS, and visionOS 27.3 or later.

2. [2]

   When sharing content stored in iCloud, users can choose “Share a Link” or “Send a Copy.” Those who choose “Share a Link” can restrict access to people they invite or approve, or allow anyone with the link to access the content.

3. [3]

   Choosing “Anyone with the link” makes the content … More → The post Apple is bringing harmful content checks to iCloud link sharing with potential account penalties appeared first on Help Net Security .

Read the original article →