The CEOs of OpenAI, Anthropic, Google DeepMind, and Microsoft AI signed a public letter on June 3, 2026, urging Congress to require synthetic DNA and RNA providers to screen every customer and every order to prevent AI-assisted bioweapon creation. The signatories warn that artificial intelligence is rapidly lowering the knowledge barrier that has historically kept biological weapons out of reach of bad actors.

Why Are AI Leaders Pushing for DNA Screening Laws?

The letter, organized by the nonpartisan Institute for Progress and the Foundation for American Innovation, argues that large language models now make it easier to identify dangerous biological sequences and design novel toxins and pathogens. While some hands-on biology training is still required to turn a digital sequence into a functioning virus, the knowledge gap has narrowed significantly.

The canonical case cited by the signatories is the 2017 horsepox reconstitution. Canadian researchers assembled the extinct virus using just $100,000 worth of mail-order DNA, and critics noted the same methodology would work on smallpox. Gene synthesis costs have continued to fall since then, making the risk more acute.

"There is a real possibility that the knowledge barriers which have historically prevented bad actors from obtaining biological weapons will meaningfully erode," the letter states.

OpenAI, Anthropic, Google DeepMind, and Microsoft AI leadership

A Microsoft study published last year showed that AI protein design tools could generate sequences with structural similarity to known dangerous proteins, and those sequences slipped past existing screening software. The current screening system matches against a library of known threats; AI systems can generate around that library entirely.

What Would the New Screening Law Require?

The ask is straightforward regulatory action: a federal mandate that any company selling printed DNA in the United States has to vet who is buying it and what sequences they are ordering. A bipartisan bill introduced in the Senate earlier this year would do exactly that, extending federal screening requirements to every provider operating in the US, regardless of whether the buyer receives federal funding.

Today, the voluntary regime runs through the International Gene Synthesis Consortium, which formed in 2009 and whose members screen orders against databases of "sequences of concern" tied to toxicity or pathogenicity. Federal guidelines from the Biden administration already require scientists and companies receiving federal funding to buy synthetic sequences only from providers that screen orders. The Senate bill would close the most obvious workaround: ordering through a non-screening vendor outside the federal funding net.

• Headline Signatories: Sam Altman (OpenAI), Dario Amodei (Anthropic), Demis Hassabis (Google DeepMind), and Mustafa Suleyman (Microsoft AI) signed the June 3, 2026 letter.
• Industry Support: Twist Bioscience and Ansa Biotechnologies, both members of the International Gene Synthesis Consortium, also signed the letter, signaling that gene synthesis companies themselves support mandatory screening.
• Defense-in-Depth Approach: The signatories argue that screening alone is insufficient; AI companies must also refuse to help users design dangerous biology at the model layer, filtering requests before they reach users.
• Competitive Advantage: A federal screening mandate paired with model-level refusals raises the compliance bar for open-weights models and smaller labs, locking the incumbents' existing safety tooling in as the de facto standard.

James Diggans, vice president of policy and biosecurity at Twist Bioscience, explained the reasoning behind industry support for stricter rules. "If you have technology that is capable of synthesizing DNA, then you should ensure that it's used responsibly, and part of that is making sure that you understand what you're making and who you're making it for," he said.

How Can AI Companies Help Prevent Bioweapon Design?

The letter pushes responsibility onto the labs that built the models, including the four whose CEOs signed it. Stanford's David Relman, who signed the letter, frames the approach as a defense-in-depth argument rather than a single fix. "Given that the screening may fail in some cases, we must then have other points of control," he explained. "That's where the AI companies are going to have to step up".

The signatories want AI companies to implement filtering at the model layer, refusing to help users design dangerous biology in the first place. This means training models to recognize requests for harmful biological sequences and declining to assist, even if the user has legitimate research credentials.

Geoff Ralston, former president of Y Combinator and partner at Safe AI Fund, summarized the principle: "It should be very difficult, if not impossible, to ask a model to help you do something imminently dangerous".

What Is the Broader Context for This Policy Push?

Bioterror attacks have historically been rare, but the failure mode the signatories are pointing at is the kind of tail risk that policymakers have struggled to price into policy. An AI-assisted design that either intentionally or accidentally seeds a pandemic represents a low-probability, high-impact scenario that the letter is attempting to force into legislation while the named labs still have leverage over how the rules are written.

For OpenAI, Anthropic, Google DeepMind, and Microsoft, signing on is also a competitive move. The companies pushing hardest for the rule are the ones best positioned to absorb its cost, a pattern that has played out in every previous wave of AI policy, from Senate testimony to executive orders on safety testing.

The letter represents a rare moment of alignment between the largest AI labs and biosecurity experts on a concrete regulatory proposal. Whether Congress acts on the bipartisan Senate bill remains to be seen, but the public commitment from four of the world's most influential AI leaders signals that the industry itself views the risk as serious enough to warrant federal action.