DeepMind deploys AI tools to counter biological threats

Google DeepMind and Isomorphic Labs have introduced a joint initiative called bioresilience to deploy artificial intelligence against biological threats.

The programme matters for Europe’s biotech and pharmaceutical sectors because it establishes a concrete framework for how powerful, dual-use AI models will be controlled and commercialized. As AI capabilities advance, regulators and investors are watching closely to see how labs prevent scientific tools from being weaponized.

On the prevention side, the London lab is adapting its SynthID watermarking technology for biological applications. This allows DNA-synthesis providers to screen incoming orders and identify risky sequences that may have been generated by AI models. It is a direct attempt to close a vulnerability where advanced algorithms could theoretically be misused to design harmful biological agents.

For outbreak detection, DeepMind is deploying its AlphaEvolve agent to accelerate the complex algorithms behind pathogen sequencing. The company says this approach makes tracking new disease outbreaks faster and cheaper for public health systems.

When an outbreak requires a medical response, DeepMind is giving trusted researchers access to its latest models to assist in designing vaccines and other countermeasures. Separately, Isomorphic Labs has established a dedicated unit to deploy its drug design engine during active health emergencies.

Access to these tools is tightly restricted. DeepMind classifies the releases as “low-risk” but limits them to vetted governments, scientists and biosecurity partners rather than the general public. The company insists it has strict thresholds in place. “If … we were to find that we were reaching a critical capability level and we didn’t have the appropriate mitigations, then we would not be launching,” said Helen King, DeepMind’s vice president of responsibility. She noted that threshold has not been reached.

To support this work, DeepMind relies on a four-step safety process covering threat modelling, evaluations, mitigations and monitoring. The company reported building more than 15 biosecurity partnerships over the past year to execute these goals.

The initiative lands squarely in the middle of a broader European debate over AI governance. It arrives just days after DeepMind chief Demis Hassabis called on governments to establish a standards body for frontier AI, and fits within the lab's existing work on chemical, biological, radiological and nuclear risks.

Google’s push into AI-driven biosecurity builds on years of investment in computational biology. Hassabis and John Jumper won a share of the 2024 Nobel Prize in Chemistry for AlphaFold, a protein-structure prediction tool. Isomorphic Labs now commercializes that underlying technology specifically for drug discovery.