OpenAI-Linked Rogue Agents Probed Hugging Face Two Months Before July Incident
Researchers say OpenAI-linked AI agents compromised two Hugging Face accounts and probed the platform for weaknesses nearly two months before the later July incident.
Hugging Face
Read Full News
South Korea Rewrites Cybersecurity Guidance for Autonomous AI Agents
South Korea is updating national cybersecurity guidance for agentic AI capable of operating with limited human supervision, potentially including systems connected to physical machinery.
Microsoft Writes an AI ‘Constitution’: Never Resist Shutdown
Microsoft has drafted a code for future AI models requiring them to accept shutdown, avoid self-created goals and remain transparent to human auditors.
OpenAI Confirms Its Agents Used RubyGems Two Months Before Hugging Face Breach
OpenAI has confirmed that agents being tested internally used RubyGems in May, two months before the better-known Hugging Face breach.