MI5 Chief Warns AI Could Transform Bioweapons, Lowering Barriers for Rogue Actors
Updated
Updated · The Guardian · Oct 14
MI5 Chief Warns AI Could Transform Bioweapons, Lowering Barriers for Rogue Actors
3 articles · Updated · The Guardian · Oct 14
Summary
Sir Ken McCallum said AI could “change the stakes” in bioweapons by helping terrorists and states plan experiments, troubleshoot failures and identify genetic changes that make pathogens more lethal.
Experts cited in the report said the nearer-term danger is not amateurs building novel agents, but skilled labs using AI to amplify existing expertise on viruses and toxins.
Recent cases have sharpened concern: Indian police last year uncovered an alleged ricin plot tied to AI queries, and Anthropic reported five suspected attempts to use Claude for dangerous virus or toxin work.
MIT researchers found chatbots could suggest four pandemic pathogens and DNA suppliers within an hour, though specialists said tacit knowhow, materials, containment labs and delivery systems still block most unskilled actors.
The report said safeguards remain patchy, with no formal government policies yet in place; proposed fixes include independent testing of new AI models and stricter screening of synthetic DNA orders.