Unquantifiable risk makes AI 'preparedness frameworks' near-worthless, says Jarovsky
LuizaJarovsky · x · 2026-10-08
AI governance researcher Luiza Jarovsky argues that if AI companies cannot reliably quantify the risks their products pose, their 'preparedness' frameworks have little to no value, and society is right to push back — a direct critique of frontier labs' risk-tiering safety documents.
More from Safety
- OpenAI's Former Safety Lead: We Now Ship 'New Capability and Risk Every Tuesday' — sourdub · 2026-10-08
- UK ARIA launches £50m Scaling Trust programme funding secure AI agent coordination — HaydnBelfield · 2026-10-08
- MIKE: researchers open-source compact post-quantum NIKE with 80-byte public keys — jedisct1 · 2026-10-08
- 'This Is Nuts': OpenAI Safety Protocol Author David Robinson Explains Why He Quit — sjgadler · 2026-10-08
- How AI surveillance in schools violates student rights and endangers vulnerable youth — nyraofficial · 2026-10-08
- Where Do You Put the Security Boundary When Agents Act on Their Own? — thefrizzybounds · 2026-10-08