DeepMind biologist who built DNA synthesizers by hand debunks AI bio-risk fear mongering

ccerrato147 · x · 2026-09-15

Anselm Levskaya, a Google DeepMind researcher credited on major transformer papers who switched from biology to AI in his 30s, pushed back hard against AI bio-risk alarmism.

He says he built DNA synthesizers and sequencers by hand and used to engineer viruses and human immune evasion for therapeutics. His core point: ordering DNA sequences from a synthesis provider is nowhere near producing a viable virus — you still must assemble the genome and bootstrap a virion production system, and viral fitness depends deeply on codons and co-translational kinetics, so a single-shot virulent design without extensive experimental selection is fantasy.

The reposting author argues the bio-risk fear mongering has no factual basis. A rare first-hand rebuttal from someone with actual wet-lab expertise in the AI bio-safety debate.

Original post →

More from AGI Musings

AGI Musings channel →