Scientists Downplay AI Bioweapon Apocalypse Risk
Experts argue that creating a bioweapon with AI faces major practical bottlenecks, despite recent alarm from tech CEOs.
AI CEOs and researchers have recently raised alarms about the potential for artificial intelligence to aid in creating biological weapons. Despite this, many scientists assess the risk of an AI-caused pandemic as relatively low compared to other global concerns.
David Bellamy, a research scientist at the Institute of Foundation Models, argues that AI does not represent a fundamentally new threat in bioweapon creation. He points to a long-standing dual-use dilemma in science regarding publishing dangerous research. Bellamy states that AI is essentially a tool that can help both good actors and threat actors find information faster. "But those capabilities are not really the bottleneck in the production of bioweapons," he adds.
The Practical Bottlenecks
The real bottleneck, according to Bellamy, is the complex physical process. A bad actor would need gene fragments, equipment, and deep knowledge to assemble a virus, verify it infects humans, causes illness, and is transmissible. Robotic lab assistants can automate some tasks, but a person still needs the expertise and resources to run the experiments.
Jason Kelly, CEO of biotech firm Ginkgo Bioworks, doubts an artificial general intelligence could commandeer labs to engineer and deploy new pathogens. He cites a recent project with OpenAI where its GPT-5 model ran a lab. "The AI could not take over the lab," Kelly says, noting humans inside could simply refuse its requests for materials. For an AGI to work autonomously, you'd have to have dramatically more robots all over the place.
Human Checks and Alternative Threats
Immunologist Derya Unutmaz agrees humans could thwart AGI attempts to develop a deadly disease. He also argues that if a supervirus were unleashed, scientists could use other AI systems to develop a vaccine rapidly.
Some experts are less worried because bioweapons are methodologically unattractive. Genetic biologist Francois Belloux notes people overestimate the value of pathogens as weapons. He says bioweapons are hard to target, complicated to mass-produce, and logistically inferior to other methods. If you want to kill people, there are much, much, much better ways, Belloux tells WIRED.
Calls for Layered Protections
Despite downplaying immediate AGI threats, experts see a need for better safeguards. Olivia Scharfman, a biotechnology fellow at the Institute for Progress, thinks AI-assisted bioterrorism is an immediate threat from certain groups. She believes this is the right moment to push for better protections against all biological threats, including upgrading building air purification.
Steph Guerra, head of AI and bio at the Rand Corporation, says it's hard to quantify if AI raises the bioweapon risk. She advocates for layered approaches. Governments could mandate screening of synthetic DNA orders. Global disease surveillance systems should be strengthened. Data-sharing protocols across AI companies, gene synthesis providers, and governments could help prevent misuse. With pretty much almost any biosecurity control, there are going to be ways that they can be circumvented, Guerra says.
Unutmaz points to a more pressing concern: that doom conversations distract from AI's positive potential in medicine. We really need to focus on the positive aspect of it, he says, such as assisting in vaccine development.
