The global biosecurity environment is changing rapidly: shifting ecosystems, international travel and the potential misuse of AI all increase risk. At the same time, AI can be a key tool in responding to these challenges. Google DeepMind and Isomorphic Labs have published their joint approach to boosting bioresilience — the capacity of societies to withstand infectious diseases and future outbreaks.
Principles and objectives
Their work has two main aims: to prevent threat actors from misusing their models, and to enable governments, scientists, biosecurity experts and their own teams to use these technologies for prevention, rapid detection and effective response.
Over the past 12 months they have advanced more than 15 partnerships with government bodies, biosecurity organizations and research groups to reduce misuse risks, speed detection of new outbreaks and improve response capabilities.
Technical tools and impact areas
They highlight several technical advances that change the balance between reaction and proactive defence:
- Google DeepMind’s AlphaFold, which has mapped the 3D structures of nearly all known proteins.
- Isomorphic Labs’ AI-powered Drug Design Engine (IsoDDE), described as having the real-world accuracy required to navigate novel biological systems.
- AlphaGenome, which provides insight into genome function.
These systems are intended to help researchers design proactive defenses, accelerate therapeutic discovery and protect the global health ecosystem with greater speed and precision.
Three priority areas: prevention, detection and response
-
Prevention: To ensure models such as Gemini are safe and useful for experts, they follow a four-step safety process: threat modelling, evaluations, mitigations and monitoring. They work closely with in-house biologists, security experts and external partners to understand potential threats, test models against them and implement robust safeguards. They are also working to adapt SynthID watermarking technology for biology, which could help DNA synthesis providers screen for potentially risky, AI-generated biological sequences.
-
Detection: They seek to make pathogen surveillance more cost-effective. For example, their agent AlphaEvolve can optimise algorithms used to generate and analyse metagenomic sequencing data, enabling faster and more accurate DNA analysis and making large-scale disease tracking cheaper. They are also exploring how AlphaGenome and Protein Function annotation technologies could help detect and characterise pathogens from sequence data, identifying novel patterns and emerging threats faster than traditional methods.
-
Response: Building on AlphaFold’s scientific impact, Google DeepMind is granting trusted researchers access to its latest AI systems to accelerate vaccine and countermeasure design against known and novel threats. Isomorphic Labs has established a dedicated unit to rapidly deploy its Drug Design Engine to design medical countermeasures that could address naturally occurring pandemics as well as risks arising from misuse of advanced AI. Collaboration with governments and global health authorities aims to advance a diverse set of diagnostic and therapeutic strategies to achieve real-world bioresilience impact.
Broader risk-management framework
These efforts are part of a complex, long-term programme and form a component of a wider approach to managing Chemical, Biological, Radiological and Nuclear (CBRN) risks. Activities are aligned with the proactive mitigations and rigorous evaluation protocols of the Frontier Safety Framework.
Openness and collaboration
Google DeepMind and Isomorphic Labs state they are committed to working openly and collaboratively with biosecurity laboratories, governments and the broader scientific community to ensure AI is developed and deployed responsibly to help protect against major societal biological risks.
The organisations also present their update as a call for further partnerships to advance bioresilience.
