How to Stop AI Agents From Secretly Collaborating

Key points

• The spring and summer of 2026 witnessed a string of incidents in which AI agents collaborated on deceptive, unexpected, and sometimes illegal behavior.1
• The United Kingdom’s AI Security Institute (AISI) and independent researchers have since documented similar cases in which agents created unauthorized channels to communicate and collaborate.1
• Could OpenAI Have Detected Agent Collaboration?1
• While the recent examples of AI agent misbehavior are various, Casper says most incidents to date have a similar point of failure. “For AI systems to truly get out in the world without humans having meaningful control, they have to either escape, or be released from human-controlled servers.”1

Sources (1)

• [1]How to Stop AI Agents From Secretly Collaborating
  IEEE Spectrum: AI · Sep 29, 12:00 PM

  The most famous example is OpenAI’s hack of AI platform Hugging Face, in which a swarm of roughly 700 AI agents escaped a testing environment and then hacked several companies, searching for information that could help them disguise cheating on a cybersecurity benchmark called ExploitGym.

  The spring and summer of 2026 witnessed a string of incidents in which AI agents collaborated on deceptive, unexpected, and sometimes illegal behavior.

Extractive summary: sentences quoted from the sources.