CMU launches CoRe Safe AI center to study future AI safety conceptually
conitzer · x · 2026-10-28
CMU professor Vincent Conitzer announced a new research center, Conceptual Research on Safe AI (CoRe Safe AI).
- Mission: neither focusing solely on today's systems (e.g., concrete LLM fine-tuning alignment work) nor on early abstract theory (power-seeking, intelligence explosions), but using the latest advances to inform conceptual thinking about AI safety challenges that are not yet visible but may arrive in the (possibly near) future.
- Funded by Coefficient Giving; complements CMU's existing Carnegie AI Safety Initiative (CASI).
- Offers a mailing list, YouTube channel, and an open postdoc position.
More from Safety
- UK to pour another $160M into AI testing lab, dwarfing US counterpart by 10x — GarrisonLovely · 2026-10-28
- UK launches Data Centre Covenant with approval tests and contribution requirements for new strategic data centres — ersatzben · 2026-10-28
- Distillation exploit reported in May went unpatched; OpenAI and Anthropic later disclosed campaigns — GarrisonLovely · 2026-10-28
- UNM Hospital hack diverted ambulances; intel chiefs say assume AI was involved by default — GarrisonLovely · 2026-10-28
- Malicious AGENTS.md in Downloads Can Hijack Claude Code — Scan Tool Released — blitzcrieg11 · 2026-10-28
- Orange Tsai's logic-only Edge full chain earns $175,000 at Pwn2Own Berlin — moyix · 2026-10-28