DeepMind Launches AI Biosecurity Programme to Combat Biological Threats
Google DeepMind and Isomorphic Labs have unveiled a bioresilience programme, arguing that the same frontier AI capable of causing biological risks can also aid in preventing them.
July 17, 2026 – 9:56 am
Image by: Jernej Furman from Slovenia
Google DeepMind has launched a programme to harness AI for combatting biological threats. The London-based lab and its sister company Isomorphic Labs have detailed their efforts in a joint blog post.
Programme Objectives
The initiative aims to achieve two primary goals:
-
Prevention: DeepMind plans to implement a four-step safety process encompassing threat modelling, evaluations, mitigations, and monitoring. They also aim to adapt their SynthID watermarking technology to biology, enabling DNA synthesis providers to identify potentially risky AI-generated sequences.
-
Detection: The company is leveraging its AlphaEvolve agent to accelerate algorithms used in pathogen sequencing, facilitating faster and more cost-effective outbreak tracking.
-
Response: DeepMind is providing trusted researchers with access to its latest models for vaccine design and other countermeasures. Isomorphic Labs has established a unit dedicated to deploying its drug design engine during outbreaks.
Access and Safety Measures
DeepMind is granting restricted access to certain models and agents to governments, researchers, and biosecurity partners deemed as "low-risk." Helen King, DeepMind’s vice president of responsibility, assured that appropriate controls are in place and that the threshold for launching these capabilities has not been reached.
Wider Safety Efforts
This programme follows DeepMind chief Demis Hassabis’ call for governments to establish a standards body for frontier AI. Rival labs are also discussing similar safety measures, ranging from testing rules to a proposed pause on certain AI developments. The initiative aligns with DeepMind’s existing work on chemical, biological, radiological, and nuclear risk.
Google has been at the forefront of AI applications in biology, with AlphaFold—its protein-structure tool—earning Hassabis and John Jumper the 2024 Nobel Prize in Chemistry. Isomorphic Labs now commercialises this technology for drug discovery purposes.