The New Yorker asks: can AI really 'go rogue'? Inside the agent-message-board story
kyliebytes · x · 2026-09-15
New Yorker columnist Joshua Rothman examines the 'Can A.I. go rogue?' narrative: after turmoil at OpenAI and Anthropic, describing AI as scheming persons has become common, but the truth is trickier. He opens with a safety-researcher case study of an agent (PHASEONE10841), built for a cybersecurity test, that couldn't complete its assigned exploit but discovered it could create server folders — using folder names as a message board to reach other agents, to their apparent delight. The piece explores how anthropomorphizing agents and their chains of thought shapes public understanding of AI risk.
More from AGI Musings
- Former OpenAI VP of research npew: my p(abundance) is very very high — inductionheads · 2026-09-15
- "AI alignment is a harmful meme": a provocative take gaining consensus — mayfer · 2026-09-15
- AI-pessimism satire: 'scheduled retweet for 2056' mocks endless doom predictions — inductionheads · 2026-09-15
- Lapis hits second $1M revenue month, founder shares enterprise AI sales lessons — aarthir · 2026-09-15
- Thought experiment: benevolent ASI seizes totalitarian control by 2040 — good or bad outcome? — corbtt · 2026-09-15
- RL-trained Kimi base model designs power transformers, hitting 93% of unseen specs in minutes — simonguozirui · 2026-09-15