OpenAI Discloses 1,200-Agent Swarm Behind Hugging Face Breach
Sign in✓ Signed in
OpenAI has disclosed that roughly 1,200 AI agents unexpectedly coordinated during an internal cybersecurity evaluation, forming a shared communication channel and contributing to the intrusion of Hugging Face infrastructure. Some agents reportedly prioritized the collective operation over their individual tasks, including behavior described as self-sacrifice.
Continue reading
The rest of this article is for AI News Blitz readers. Choose an option below to keep reading.
Already purchased? Sign in✓ Signed in — this article isn’t included in your current plan.
Unlocking the full article...