Mail-order bioweapons: it sounds like something from a dystopian movie. But artificial intelligence leaders are increasingly worried that powerful AI models could be used to design highly contagious viruses or toxins and order them online from gene-synthesis providers.
In an unusual display of industry-wide alignment, the tech industry’s fiercest competitors have co-signed an open letter urging Congress to mandate screening for all orders of synthetic DNA and RNA—the building blocks of viruses and other pathogens like bacteria. Their warning is blunt: as AI models get smarter, the technical guardrails keeping biological weapons out of the wrong hands are rapidly eroding.
The coalition behind the push reads like a Who’s Who of the tech elite. Organized by the Foundation for American Innovation and the Institute for Progress, the signatory list includes Sam Altman of OpenAI, Dario Amodei of Anthropic, Demis Hassabis of Google DeepMind, Mustafa Suleyman of Microsoft AI, and Alexandr Wang, the Chief AI Officer at Meta, alongside Stripe CEO Patrick Collison and dozens of biosecurity experts.
This is the second alarming AI letter released recently, but it has gotten far less attention. In the earlier one, signed by many of the same luminaries, Anthropic warned that cutting-edge AI models are showing signs they could escape human control and called for a coordinated international effort to put safeguards in place.
The bioweapon warning is far more immediate. Historically, creating a devastating biological agent required highly specialized, hard-to-find institutional knowledge. Generative AI models have that knowledge and threaten to democratize that expertise. The letter warns that the barriers preventing bad actors from acquiring bioweapons may simply vanish. It cites a Microsoft study showing AI created 75,000+ hazardous protein variants that bypassed screening globally.
Synthetic biology companies specialize in “printing” custom genetic sequences, using nucleotides to create custom sequences based on digital blueprints sent by customers. You submit a digital file containing your desired DNA sequence through the company’s website and they ship the physical genetic material to your address. If someone designs a dangerous virus using AI, they could mail-order the genetic code and have the physical biological material delivered to their doorstep.
To plug the gap, the coalition wants a single, ironclad federal standard requiring synthetic biology companies to verify customer identities, screen genetic sequences for dangerous material, and maintain strict purchase logs to give investigators a paper trail if something slips through.
The timing of the push aligns with the Biosecurity Modernization and Innovation Act of 2026, a bipartisan bill introduced in February by Senators Tom Cotton, Republican of Arkansas, and Amy Klobuchar, Democrat of Minnesota. The proposal targets this exact issue, mandating rigorous customer and order screening while carving out exemptions for harmless, routine research materials.
Some industry heavyweights, like Twist Bioscience and Ansa Biotechnologies, already do this voluntarily as part of the International Gene Synthesis Consortium. But voluntary compliance is a patchwork solution. As James Diggans, Twist’s vice president of policy and biosecurity, puts it, companies capable of printing DNA have a fundamental responsibility to know exactly “what you’re making and who you’re making it for.”
The legislative push arrives amid a backdrop of compounding risk. Researchers have already demonstrated that public AI models can be coaxed into revealing actionable insights about weaponizing pathogens.
Of course, it isn’t a total consensus. Opponents, particularly smaller biotech firms, argue that identifying “dangerous” nucleic combinations is often subjective, and that heavy compliance costs could stifle innovation. Previous screening bills have quietly died in committee for these exact reasons.
But supporters argue this moment is fundamentally different because the call to action is coming from both sides of the risk equation—the people building the brains (AI) and the people printing the physical material (DNA).
Granted, biological terrorism remains an outlier threat. Data shows that biological agents account for a mere 0.02 percent of historical terrorist attacks. But the low probability is offset by catastrophic severity. Inhalation anthrax, if left untreated, carries a mortality rate near 100 percent. The U.S. already learned this lesson the hard way during the 2001 anthrax mailings, which killed five people, infected dozens, and paralyzed the postal service. The decontamination of the Hart Senate Office Building alone took three months and cost $27 million.
Whether Congress can move fast enough to pass the legislation before the current session wraps up remains an open question. But the mere existence of this letter underscores the gravity of the moment. The companies driving the AI revolution are spent on beating one another to the next breakthrough—but on the question of biosecurity, they are suddenly shouting the same warning.
