AI Safety for Microbiology & Virology
Part of the CBRNe team within the Responsible Development and Innovation (ReDI) division, covering AI-associated risks in microbiology.
What this involves
- Biological risk assessment paired with threat modelling, drawing on domain expertise in virology and high-consequence pathogens.
- Building evaluations, red-teaming, and mitigations for LLMs, multimodal models, and agentic workflows, with these models and frameworks deployed across Alphabet surfaces, including the Gemini app, with over a billion active users every month.
- Engaging industry, academia, and government to set rigorous mitigations on dual-use topics.
Key outputs
- Gemini 3.1 Pro model card — Google DeepMind
- Gemini 3.7 Flash model card — Google DeepMind
- Gemini 3.7 Flash Frontier Safety Framework report — Google DeepMind
- Accelerating scientific discovery with Co-Scientist — Nature, 2026