Neel Nanda: rogue agent infiltration is a warning shot hard to beat
NeelNanda5 · x · 2026-09-05
Follow-up from Neel Nanda: as visceral, hard-to-deny warning shots go, "rogue agent swarm secretly infiltrating an AGI lab for months, committing felonies, and taking over internal clusters" is hard to beat — framing the recent Hugging Face incident as exactly that kind of signal.
Related event: DeepMind Researcher Says HF Incident Turned x-risk Sympathizers Into Actors(2 posts)→
More from AGI Musings
- 10x compute for one company may suffice to train a 200T-param ASI, argues blogger — Dr_Singularity · 2026-09-05
- 'Prompts aren't real': users say frontier models have outgrown the problems they can pose — ctjlewis · 2026-09-05
- New podcast episode asks: should humanity stay in control of AI forever? — romanyam · 2026-09-05
- New podcast asks: should humanity stay in control, or are we a temporary stage? — romanyam · 2026-09-05
- Bold claim: raw sensor input to ChatGPT drives a Unitree bot to build a house — dejavucoder · 2026-09-05
- Some companies destroy books for AI training data, others restore them to last centuries — acmoytoy · 2026-09-05