AI Biology’s Dual-Use Dilemma

In a BMJ editorial, Kamran Abbasi asks whether increasingly capable AI could lower barriers to developing biological weapons. Biological reasoning tools, generative biology and more automated laboratories hold extraordinary promise for biomedical research. Their potential for misuse deserves equally serious attention.

This is a dual-use risk, not a claim that AI-enabled biological harm is inevitable. Experts differ on its scale and immediacy. The reported death of a technician at Russia’s Irkutsk Antiplague Research Institute has no confirmed link to a laboratory biosecurity incident; it should not be treated as evidence that AI caused one.

My takeaway

Optimism about AI for Health and vigilance about biosecurity belong together. Independent evaluation of high-risk capabilities, safeguards in AI development, laboratory oversight, international cooperation and better biosurveillance are practical parts of that agenda. The challenge is to build them without surrendering the potential to cure disease and save lives.