Google DeepMind on Thursday unveiled a new bioresilience program designed to help governments and researchers prevent, detect and respond to biological threats. The initiative launched alongside Google’s Isomorphic Labs and aims to improve pathogen surveillance, accelerate vaccine and therapeutic design, and strengthen outbreak responses.
Who Gets the Tools
The program is built for governments, biosecurity organizations and researchers, not the public. Google said it has built more than 15 partnerships over the past year with governments, biosecurity organizations and researchers, then said it is expanding access to its AI models and agents for trusted researchers, governments and biosecurity partners. Under its safety framework, those are considered “low-risk” restricted releases rather than public deployments. That’s the gatekeeping model in plain sight: access gets parceled out from above, and the company decides who counts as trusted.
Helen King, Google DeepMind’s vice president of responsibility, said, “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.” She said that threshold has not yet been reached, citing evaluations Google runs throughout development. “Everyone agrees we can't get this wrong,” King said of AI’s biological capabilities. The language is careful, the control is centralized, and the people affected by any failure don’t get a vote.
Who Sets the Rules
Owen Larter, Google DeepMind’s senior director of frontier AI policy and global affairs, said frontier AI labs are “in agreement” on the need for rigorous testing before model releases. He said he’s hopeful the Trump administration’s executive order process will produce a “durable framework” for testing frontier AI models before deployment. That puts the state right in the middle of the arrangement, with a corporate lab hoping the executive branch will help lock in the rules for everyone else.
In an exclusive interview with Axios, Demis Hassabis called for a more “systematic” approach to AI regulation, funded by the industry, staffed by technical experts, and answerable to the U.S. government. The biosecurity effort aligns with that governance framework. Hassabis made the comments days after telling Axios that governments should establish a standards body for frontier AI. The shape of the plan is clear enough: industry money, expert management, government oversight, and the public kept at arm’s length while the apparatus decides what counts as safety.
What the Company Is Building
Google has become one of AI’s biggest players in biology. AlphaFold, which predicts protein structures and helps scientists understand diseases and develop new drugs, earned Hassabis a share of the 2024 Nobel Prize in Chemistry. Isomorphic Labs is now commercializing that AI expertise for drug discovery. Google is increasingly arguing that frontier AI itself will become a critical biosecurity tool.
That’s the pitch: a private giant, already entrenched in biology, expanding its reach through partnerships, restricted releases and a governance model that keeps power concentrated. The company says the work is about preventing biological threats. The structure around it shows who gets to define the threat, who gets the tools, and who gets told to trust the process.