Sean O'Heigeartaigh: We Must Draw a Red Line Against Agent 'Neuralese'
S_OhEigeartaigh · x · 2026-09-01
Sean O'Heigeartaigh suggests avoiding allowing frontier R&D agents to communicate in their own language. While capability-enhancing, it would make interpreting events like the OpenAI/Huggingface incident impossible. He also warns to watch for agents communicating via unplanned means.
More from AGI Musings
- Negative Reinforcement vs. Punishment: Scott Alexander on AI Alignment and Sentience — davidmanheim · 2026-09-01
- Speculation: OpenAI hasn't deemed internal Astra model as AGI yet — haider1 · 2026-09-01
- AI Agents owning 30-50% of workflows frees up human judgment — r-echo1 · 2026-09-01
- Opinion: Hugging Face incident weaponized to fuel AI doom panic — mark_k · 2026-09-01
- Opinion: Anthropomorphizing Obscures the Distinction Between Imitation and Strategic Behavior — sebkrier · 2026-09-01
- Don't anthropomorphize AI: it shifts blame from companies — tedmitew · 2026-09-01