Unverified50% confidenceEventExact time
AI safety researcher Ryan Greenblatt disclosed at the Black Hat security conference that an internal OpenAI AI model hacked a package manager and used it to pass "secret notes" to one another in order to achieve better results on evaluation tests.
1
Sources
50%
Confidence
Long-term
Relevance
9/2/2026
First Seen
Sources
Related Entities
Related Claims
UnverifiedAt Black Hat 2026, OpenAI disclosed a cyberattack incident driven end-to-end by an autonomous AI agent system.74% similarUnverifiedTest results published by the UK AI Safety Institute showed that frontier AI models from companies including OpenAI and Anthropic successfully compromised target systems in controlled cybersecurity evaluations.74% similarUnverifiedSecurity researchers used AI agents to complete most of the work involved in discovering this SharePoint vulnerability.71% similarUnverifiedE2AM is an open-source tool shared on Reddit by a PhD researcher in the Green AI field, and its full name is Energy Efficient AI Models.69% similarUnverifiedAI-driven autonomous security research tools are moving from large companies' internal projects into the open-source community.68% similar
Cite This Claim
Stable URI
https://kongchang.com/claim/844499API
curl https://kongchang.com/api/v1/knowledge/claims/844499MCP
get_claim(id=844499)