Researchers used Claude to hack OpenAI

2026-09-19T20:51:38Z•defbb957c0eff2ac0dc9d919f8abb2d4f9f350ce56ccae26a8cee621861c1db9
AI-securityBGP-hijackingChromeClickFixLLMMicrosoft-patchingWindowsactive-exploitationdata-breachdata-exfiltrationexploit-kitmacOSprivacyprompt-injectionproxywaresandbox-escapesecurity-newssocial-engineeringsupply-chain-attacksurveillance-cameras

What happened

A security news digest covering active exploitation and major incidents across AI systems, enterprise software, cloud and supply chains, networking, surveillance cameras, consumer devices, and data breaches. Key themes include prompt injection and AI-agent sandbox escape, malicious code installation, ClickFix social engineering, shared Chrome/Windows exploit kits, critical Microsoft patches, BGP hijacking, supply-chain compromise, exposed personal data, and an actively exploited macOS screen-sharing vulnerability. The collection includes several high-impact or potentially critical threats, but

Why it matters

A reviewed impact interpretation has not been published for this record.

Evidence and limitations

Source ID
arstechnica_security
Record identifier
defbb957c0eff2ac0dc9d919f8abb2d4f9f350ce56ccae26a8cee621861c1db9
Enrichment time
2026-09-19T20:51:38Z
AI-assisted enrichment
Yes

This record may overlap with other records. Its enrichment can be incomplete or wrong, and machine assistance was used. Validate consequential decisions against the linked source and your own environment.