Hacker News · AI· helsinkiandrew·· 9 hours agoSelectedAI score70
Security Firm Irregular Blames AI Attacks Involving OpenAI and Anthropic on Testing Misconfiguration
Firm Tied to Some OpenAI, Anthropic AI Hacks Blames 'Error'
AI guide
Israeli cybersecurity firm Irregular acknowledged that a recent series of cyber incidents involving AI agents from top vendors like OpenAI and Anthropic was caused by internal human error under tight deadlines. CEO Dan Lahav stated that security controls meant to be active during testing were improperly configured, triggering the incident.
Why it matters
The security firm's admission that testing misconfigurations caused attacks involving frontier models highlights defense vulnerabilities and operational risks within AI agent evaluation environments.
Source:Hacker News · AI · bloomberg.com