The CEOs of some of the most powerful AI companies in the world are asking Congress to act before their technology gets used to help someone build a biological weapon. OpenAI, Anthropic, and other leading labs have signed an open letter urging lawmakers to strengthen rules around synthetic DNA sequences, which can now be ordered online and, in the wrong hands, could be used to engineer dangerous pathogens.
The letter arrives at a moment when AI systems are becoming genuinely useful for scientific research, including biology. That is mostly a good thing, but it also means the same tools that help researchers understand diseases can potentially help bad actors design them. The gap between what a trained scientist and an untrained one can do with AI assistance is narrowing, and policymakers have been slow to respond.
The core ask is straightforward: tighten screening requirements for companies that synthesize and sell DNA. Current rules in this area are inconsistent and, in some cases, voluntary. The signatories want Congress to make robust screening mandatory across the board.
Why synthetic DNA screening matters
Synthetic DNA is the raw material for a lot of modern biology. Researchers routinely order custom genetic sequences from commercial suppliers to conduct experiments. But the same supply chain that serves legitimate science can also serve someone trying to reconstruct a dangerous virus or enhance a pathogen’s ability to spread.
Right now, screening practices vary widely between suppliers. Some run sequences against databases of known dangerous pathogens before fulfilling orders. Others do not. The letter argues that AI tools make this gap more dangerous than it used to be, because AI can help users design sequences that partially evade existing screening methods or guide someone through technical steps they would previously have needed expert knowledge to complete.
Who signed and what they’re asking for
The letter includes signatures from AI lab executives as well as biosecurity researchers and scientists. While the full list has not been published in detail, the involvement of OpenAI and Anthropic, the two most prominent names in commercial AI right now, gives the effort significant visibility on Capitol Hill.
Their specific requests to lawmakers include:
- Making DNA synthesis screening mandatory for all commercial suppliers, not just those who opt in
- Requiring suppliers to check orders against comprehensive, up-to-date databases of dangerous sequences
- Funding research into better screening tools that can keep pace with advances in AI-assisted sequence design
- Creating a federal framework that coordinates oversight across agencies
The broader context: AI and biosecurity
This letter is not the first time the AI industry has raised alarms about biological risks. Earlier studies, including research conducted by Anthropic and others, have tested whether large language models could provide meaningful “uplift” to someone trying to build a bioweapon. The results have been mixed but concerning enough that several labs have made biosecurity a specific focus of their safety work.
The Biden administration previously issued an executive order that touched on AI and biosecurity, and the current administration has signaled interest in maintaining some of those guardrails. Congressional action, though, has been limited. A mandatory screening law for DNA synthesis has been discussed for years in biosecurity circles but has not passed.
Whether this letter changes that calculus is unclear. But having the AI industry itself call for regulation, rather than resist it, is a notable shift. These companies are essentially telling Congress that the technology they are building creates risks that voluntary industry action alone cannot manage.
What happens next
The letter is addressed to members of Congress, though specific committee targets have not been confirmed publicly. Biosecurity advocates have long argued that the window to get ahead of this problem is closing as AI capabilities improve, and that reactive legislation passed after an incident would be far less effective than rules put in place now.
For the AI companies involved, signing on also carries a practical benefit: it signals to regulators and the public that they are thinking seriously about misuse, not just capabilities. That kind of visible engagement with hard safety questions matters as Congress debates broader AI legislation.