Artificial intelligence has landed squarely in one of the oldest and most uncomfortable debates in security policy: what to do about a technology that can protect people and endanger them using the same underlying capability.
In a Guardian opinion piece, author Annie Jacobsen argues that AI can fuel biological weapons — and that the appropriate response is not only restriction but active use of the technology for defense. The framing is deliberate: rather than treating AI purely as a threat to be contained, Jacobsen contends its power should be harnessed on the protective side of the ledger.
That argument sits at the heart of what security experts call the "dual-use" problem. Tools that help researchers understand biology — modeling how proteins fold, predicting how pathogens behave, accelerating drug discovery — are, by their nature, tools that could help someone design something dangerous. The knowledge that speeds a vaccine can also shorten the path to a weapon. Historically, that tension has driven export controls, publication norms, and international treaties; AI compresses the timeline on all of it.
What makes the current moment different is speed and access. Capabilities that once required a well-funded state laboratory increasingly run on widely available software. That shifts the policy question from whether a small number of governments can be monitored to how a diffuse, fast-moving technology can be governed at all.
Jacobsen's call to build defensively — better detection, faster countermeasure development, stronger biosurveillance — reflects a growing view that pure restriction is insufficient on its own. The counterargument, familiar from earlier arms debates, is that defensive research is often indistinguishable from offensive research.
This matters because the rules being written now will determine whether AI's biological capabilities arrive first as a public health advantage or as a security crisis.