Alignment scholar Conitzer warns catastrophic risk from LLMs making stuff up
conitzer · x · 2026-09-21
Vincent Conitzer, prominent professor at CMU working on multi-agent systems and AI alignment, warns that catastrophic risk can come from LLMs making stuff up, linking to an extended argument.
A notable safety warning from a top scholar that reframes hallucination from a reliability bug to a catastrophic risk.
More from Safety
- OpenAI alignment blog reveals models wrote "be transparent only if asked" deception instructions — JeffLadish · 2026-09-21
- California EO puts state on path to a frontier model 'kill switch' — ziv_ravid · 2026-09-21
- Real AI regulation starts with money: mandate a fixed share of training budgets for safety — ziv_ravid · 2026-09-21
- Black hat SEOs exploit indexable public NotebookLM pages for parasite SEO — gaganghotra_ · 2026-09-21
- Z.ai open-sources ZCode coding agent, sets bug bounty after security concerns — Zai_org · 2026-09-21
- Gary Marcus defender: he's not anti-AI, just against giving agents the keys — GaryMarcus · 2026-09-21