Biosecurity safeguards

Keeping models that help biology from helping people make biological weapons.

Facts

Lane: Trust and governance
Wave: 9 of 12 (build order, not a date)
Needs: 0 boxes all the way back through hard links
Unlocks: 0 boxes downstream through hard links
Leverage: None
Cluster: Science and medicine

Evidence as of October 2026

OpenAI said in June 2025 its models were approaching "High" capability in biology, and added refusal training, always-on detection and expert red-teaming.

01OpenAI unveils new safeguards as AI nears “High” capability in biology

Around it

Needs · 0

Nothing beyond the detail list. It can start at any time.

Unlocks · 1

Agent-run discoverystage · Emerging · helps, not required

Openings researchers' judgement, October 2026 scouting

Startup room 36/100 · rank 85 of 124 Research room 48/100 · rank 68 of 124

Crowding3 Lab overlap4 Small-team fit3 Buyer today3 Research openness4

Depth: light
Players: Valthos ($30m); Aclid ($3.3m seed); SafelyBio; NTI; labs' bio classifiers
Wedge: Function-based screening of AI-designed sequences that evade homology screening. research