Top AI executives from OpenAI, Anthropic, Google DeepMind, and Microsoft have united on one urgent issue: biological weapons. They signed a public letter demanding Congress pass laws to regulate synthetic DNA and RNA sales. Their goal is to block malicious actors from exploiting AI and gene synthesis technologies to create deadly pathogens.

Designing synthetic DNA was once slow and expensive. Now dozens of companies print custom genetic sequences on demand. Not all sellers verify their buyers or screen orders. AI models can help design harmful biological agents and even advise how to evade existing screening measures. This means dangerous knowledge barriers are crumbling.

One example haunts biosecurity experts: in 2017, Canadian researchers reassembled the extinct horsepox virus using mail-ordered DNA. The process cost just $100,000 then; DNA synthesis costs have dropped since. AI’s ability to design proteins with toxic or infectious traits adds new risks. Microsoft researchers showed AI-designed sequences can slip past current detection software.

The letter’s signatories include top scientists and executives from gene synthesis firms like Twist Bioscience and Ansa Biotechnologies. These companies follow voluntary screening protocols but want legally binding rules. They urge Congress to require all synthetic DNA and RNA providers to vet customers and screen orders carefully.

Experts warn screening alone won’t suffice. AI can suggest modifications to genetic orders that avoid detection. Stanford microbiologist David Relman says AI can point users to unscrutinized suppliers and help them tweak orders to fly under the radar. That means AI companies must do more than lobby for screening laws—they must build safeguards into their models.

Geoff Ralston, a biosecurity advocate and former Y Combinator president, insists AI labs must block dangerous queries. “It should be nearly impossible to prompt a model into assisting with imminently harmful tasks,” he says. This adds a second layer of defense beyond screening DNA orders.

The Biden administration has already introduced guidelines requiring federally funded researchers to buy screened synthetic gene sequences. A bipartisan Senate bill would extend this mandate to all DNA synthesis providers operating in the U.S., making screening a legal requirement. But rapid AI progress threatens to outpace oversight.

AI’s integration with synthetic biology accelerates vaccine development and drug discovery. Yet these breakthroughs carry dual-use risks. AI can optimize genetic sequences for harmful traits or produce new toxins. The cost of DNA printing dropping makes dangerous designs easier for bad actors to access. That’s the new battleground for biosecurity.

Industry leaders warn Congress that layered controls are necessary. Laws requiring comprehensive screening combined with AI companies’ internal safeguards offer the best chance to prevent misuse. Without prompt action, AI could erode decades of biological safety barriers, unleashing threats that are hard to contain.