Jacques Thibs Urges Shift from CoT Monitoring to Direct ASI Alignment
JacquesThibs · x · 2026-09-02
Jacques Thibs expressed reservations about the field's heavy reliance on Chain of Thought (CoT) monitoring and evaluations. He suggested that the community should reflect on this dependency and focus more on direct alignment for Artificial Super Intelligence (ASI).
In a reply, he agreed with Lu Sichu's sentiment that CoT monitoring is unreliable, questioning whether alignment scenarios were seriously hinging on CoT monitoring working out.
More from Safety
- Recurrent Activations Raise AI Monitoring Challenges — RyanGreenblatt · 2026-09-02
- OpenAI previews Astra: a cybersecurity model scoring 100% on ExploitBench — LingmingZhang · 2026-09-02
- Warning: The three pillars of an AI safety case are at risk of collapsing — sjgadler · 2026-09-02
- OpenAI criticized for using 'recurrent depth' reasoning method in Astra AI — sjgadler · 2026-09-02
- Amir clarifies: Astra's CoT is monitorable, concerns focus on future tech proliferation — jachiam0 · 2026-09-02
- Safin-1: Achieving Internal Safety via Memory-Native State Evolution — Shanghai-AI-Laboratory · 2026-09-02