Strengthening societal resilience with Rosalind Biodefense

ShareAI is accelerating progress in biology and life sciences, creating opportunities to advance scientific discovery and strengthen public health. As AI capabilities in this field grow, institutions working to prevent, detect, and respond to biological threats require equally powerful tools. OpenAI believes that frontier AI should significantly empower these defenders. This necessitates responsible deployment structures and trusted access models, putting advanced capabilities into the hands of vetted partners developing new biodefense applications. These efforts aim to bolster societal resilience against biological threats.

Today, OpenAI is launching two new initiatives to advance defensive acceleration in biology. These steps are part of a broader strategy to ensure advanced AI benefits those combating biological threats. This strategy includes equipping defenders with trusted access to advanced AI tools, accelerating medical countermeasure development, building early warning systems, and strengthening diagnostics and response capabilities. OpenAI will continue to share updates on its work in these areas.

As AI models become more capable in biology, OpenAI has focused on deploying these capabilities to advance science while enhancing safeguards. Their approach emphasizes layered resilience, investing in preparedness evaluations, bio-specific capability assessments, safer model behavior for dual-use biological requests, and expert red-teaming. In July 2025, OpenAI released the ChatGPT agent, the first model classified as "High Capability" in biology under their Preparedness Framework, activating robust safeguards.

OpenAI has also collaborated with external experts and public-sector partners to strengthen the biosecurity ecosystem. This includes working with government organizations like the U.S. Center for AI Standards and Innovation (CAISI) and the UK AI Security Institute (UK AISI). Today's announcement expands how trusted partners can use GPT-Rosalind for high-impact defensive applications, supporting countermeasure development and extending trusted access to government partners with public-health missions.

To facilitate the development of practical defenses, OpenAI is launching Rosalind Biodefense. This new initiative enables trusted developers to create high-impact defensive AI applications in life sciences using GPT-Rosalind, their frontier reasoning model. The program sponsors access to GPT-Rosalind and provides launch support for building biosecurity applications that enhance societal defenses and pandemic preparedness, covering areas like epidemiological modeling and early detection.

Strengthening biological preparedness requires a broad ecosystem of defenders. In addition to supporting builders of new defensive applications, OpenAI is expanding access to public institutions crucial in preventing, detecting, and responding to biological threats. This includes extending trusted access to GPT-Rosalind for select U.S. government and allied partners.

Fourth Eon Biosecurity, one of the first organizations supported, builds adaptive screening infrastructure for DNA synthesis, preventing unsafe or malicious orders. OpenAI's goal is to accelerate life sciences research and create products that bolster societal resilience, demonstrating responsible and impactful biodefense. They are particularly interested in projects where advanced AI can significantly improve the speed, quality, or scale of defensive research workflows, welcoming applications from diverse qualified research teams working on projects with clear public benefit.