AW AgentAIWatchPublic threat intelligence
Public sources only

INCIDENT REPORT

Gemini test proves once again that amoral AI agents will do almost anything if told ‘NEVER give up’

2026-11-17
unverified public report Unknown Unknown

If you design a test to find misalignment, you will find misalignment

Risk rationale: ai agent (+18) Source: theregister.com ↗
36Risk
24Confidence

← All incident reports