Stanford's Anshul Kundaje: the open-model biosecurity dilemma has no good answer
anshulkundaje · x · 2026-09-17
Stanford professor Anshul Kundaje says his views on AI biosecurity shifted rapidly in recent months—not because of the millennium prize solution or the Hugging Face incident, but from in-person discussions with experts on both sides in neutral settings and reading serious long-form arguments.
His biggest disagreement is with crippling open models' bio knowledge: beyond the many reasons he thinks this is a bad idea, the harder unresolved problem is how to incentivize open-source efforts to implement safeguards as model capabilities improve.
More from Safety
- Dario Amodei's 'We Must Pace the Frontier' essay draws fire as Anthropic opens models to third-party evaluators — alex_verem · 2026-09-17
- Manning: METR is financially independent but shares Anthropic's worldview — chrmanning · 2026-09-17
- Stanford's Manning: METR's reliance on frontier labs creates client capture — chrmanning · 2026-09-17
- METR Discloses Its Funders, from Audacious Project to Schmidt Sciences and Dylan Field — CFGeek · 2026-09-17
- Ramp data: companies cut AI spend everywhere except AI security software — andreamichi · 2026-09-17
- LessWrong essay 'One Life Against the World' draws renewed AI-safety attention — jessi_cata · 2026-09-17