A September 15, 2026, report said OpenAI backed three bipartisan U.S. bills addressing biological data and AI-related biosecurity. The proposals would assign separate roles to the National Institute of Standards and Technology (NIST) and the U.S. Department of Energy (DOE). Congress.gov lists all three as introduced.
What the three bills would do
The SCALE Biology Act and the AI-Ready Bio-Data Standards Act focus on NIST; the Web of Biological Data Act would put DOE in charge of a centralized resource for biological data. Their proposed functions differ:
| Bill | Identifier | Proposed role |
| SCALE Biology Act | H.R.8981 | Establish a NIST measurement-research program for engineering biology and biological measurement, with standards and guidance related to secure biotechnology development. |
| AI-Ready Bio-Data Standards Act | S.4069 | Direct NIST to develop definitions, standards, resources and frameworks for biological datasets, with the aim of making federally funded data usable for AI research. |
| Web of Biological Data Act | H.R.9307 | Direct the Energy Secretary to establish a centralized resource for accessing biological data. |
In practical terms, two proposals center on NIST’s measurement and data-standards work. The third focuses on centralized access to biological data through DOE. Congress.gov lists each measure as introduced, not enacted.
OpenAI’s stated rationale: risk evaluation and defense
Richard Johnson, OpenAI’s national security risk mitigation lead, described the company’s approach as addressing biosecurity risks while also considering defensive uses of AI. He said measuring capabilities and risk matters, as does putting useful tools in defenders’ hands.
That dual focus is reflected in the proposals’ different functions: measurement and standards on one side, and access to biological data on the other. The AI-Ready Bio-Data Standards Act, in particular, would seek to make federally funded biological datasets usable for AI researchers.
The biosecurity context
The legislative push comes amid concerns about AI-assisted biological research. Researchers from Stanford University and the Arc Institute were reported to have designed 16 new viruses with AI assistance. Separately, Anthropic disclosed five cases in 2026 involving foreign actors’ illicit use of its models for dangerous pathogen research that could support bioweapon development.
Biosecurity expert Ashish Jha argued that safeguards can block harmful requests on one AI model, while users may switch to a less restricted model. Controls on an individual system, in other words, may not address every route a user could take.
