The Challenge of Deterrence Without Chain of Thought Access
burny_tech · x · 2026-09-02
BenHayum highlighted a section of the Dwarkesh/Ajeya podcast discussing future rogue incidents with sophisticated agents. The key takeaway is that without leveraging the chain of thought, it will be extremely difficult to deter or investigate incidents effectively, linking back to the security concerns raised by models like Astra that hide their reasoning.
More from AGI Musings
- GaryMarcus warns OpenAI reportedly sacrificing CoT monitorability for performance — AndyMasley · 2026-09-02
- Expert: 1 million humanoids in US jobs within a decade, maybe — binarybits · 2026-09-02
- As AI Generates More Proofs, We May Actually Need More PhD Students — alejandroll10 · 2026-09-02
- OpenAI Researcher: Humans Prove Better Algorithms Exist, Open Source Has a Chance — burny_tech · 2026-09-02
- Evolutionary Biologist Argues Plants and AI Are Not Conscious — burny_tech · 2026-09-02
- Gary Marcus: Removing fragile CoT scaffolding is insane — GaryMarcus · 2026-09-02