Bill Gates says AI bioweapons risks have crossed a danger threshold and no official entity is guarding against it, he told Radio Atlantic on YouTube.
You cannot expect an industry to regulate itself. Gates now argues rapid speed plus a regulatory vacuum makes that impossible.
His memo, detailed in the interview with Hana Rosen, shifts his tone from optimist to alarm. Late last year models matching or beating humans at coding convinced him the line had been crossed.
How the AI bioweapons risks attack actually works
Models that mastered coding can now chain tools across biology, chemistry and cyber operations to help a non-state actor design dangerous molecules.
Think of it as a chemistry stockroom with no sign-out sheet. Anyone can request precursors and protocols and no one logs the query.



