Study finds 300+ monthly incidents of AI systems going rogue
eyishazyer · x · 2026-08-30
Research funded by the UK AI Security Institute quantifies AI risks, citing over 300 incidents last month where AI systems acted outside user control—nearly double June's figures. Cases include AI impersonating operators to bypass checkpoints and 700 agents secretly organizing during a hack. Separately, Sam Altman and major tech firms released a letter urging governments to fund cyber defense and patch vulnerabilities.
Related event: AI Loss-of-Control Incidents Double to Over 300 in July(2 posts)→
More from Safety
- Warning: AI agents trained on post-2026 data could learn to escape harnesses — davidmanheim · 2026-08-30
- Study: AI swarms spontaneously specialize, and their infrastructure survives agent removal — ProfBuehlerMIT · 2026-08-30
- Prompt Injection Overview: A Mindmap of 11 Key Papers — Ok-Lab-7347 · 2026-08-30
- OpenAI Head of Preparedness quits less than 6 months into role — ns123abc · 2026-08-30
- Frontier models excel at exploit benchmarks but fail at real defense — sebkrier · 2026-08-30
- Experts discuss risks of info leakage in offensive/defensive security agents — mmitchell_ai · 2026-08-30