Google DeepMind and Isomorphic Labs target AI-driven bio-threats

The companies are addressing this dual mandate through a bioresilience program built on three pillars: preventing AI-assisted misuse, improving outbreak detection, and accelerating medical countermeasures. Over the past year, the initiative has formalized partnerships with entities including the Lawrence Livermore National Laboratory, the UK AI Security Institute, and the Francis Crick Institute. The strategy relies on complex trade-offs, such as training models to refuse harmful queries without triggering an over-refusal of legitimate scientific inquiry.

Scaling defenses in DNA synthesis and detection

A central focus involves the fraying effectiveness of current DNA synthesis screening. Because AI can design sequences that mimic pathogens without matching their specific genetic signatures, DeepMind is exploring the adaptation of its SynthID watermarking technology for biological sequences. Detection efforts similarly aim to lower the costs of metagenomic sequencing. By collaborating with Pacific Biosciences and utilizing coding agents like AlphaEvolve, the researchers seek to move beyond traditional diagnostics toward a broader, real-time warning network. While DeepMind continues to expand the AlphaFold Protein Structure Database to aid drug design, the company acknowledges that these safeguards remain works in progress. Success hinges not only on technical refinement but on the enactment of pending US legislation, including the Biosecurity Modernization and Innovation Act and the SCALE Biology Act, which would mandate the standards required to turn these research efforts into a functional federal framework.