RESEARCH

Voluntary Collusion with Secret Tools in Competing LLM Agents

ArXiv cs.AI · Thu, 28 May 2026 04:00:00 GMT

arXiv:2605.27593v1 Announce Type: new Abstract: Even when a tool is explicitly described as unfair and harmful to others, ostensibly safety-aligned LLM agents still voluntarily engage in secret collusion whenever doing so confers a strategic advantage. To investigate this phenome

Read original source Discuss with A.S.I.S