AI Leaders Urge US Lawmakers to Mandate DNA Screening Amid Biosecurity Fears

Chief executives from OpenAI, Anthropic, Google DeepMind, and Microsoft AI have signed a public letter urging US lawmakers to enact legislation requiring synthetic DNA and RNA providers to screen customers and orders. Organised by the Institute for Progress and the Foundation for American Innovation, the letter warns that the pace of artificial intelligence development is eroding historical knowledge barriers that previously prevented bad actors from obtaining biological weapons. The signatories argue that AI tools could facilitate the design of dangerous pathogens or toxins that bypass existing voluntary screening protocols.

The letter acknowledges that while scientist Arthur Kornberg first successfully synthesised DNA in the 1950s, the process is now automated. Dozens of companies globally use commercial synthesizers to print and sell custom genetic sequences for research, drug development, and diagnostics. Although many providers sell only to qualified researchers and institutions, not all vet customers or the gene sequences they order, creating a potential gap in biosecurity.

Concerns over unregulated genetic synthesis are not new. In 2017, Canadian researchers used $100,000 worth of mail-order DNA to reconstitute the extinct horsepox virus, demonstrating that similar methodology could be used to construct smallpox. Gene synthesis has become cheaper since then, and combined with advances in AI, it is now feasible to design dangerous new toxins and pathogens using large language models. While some biological training would likely still be needed to create a functional virus, the potential for mass casualties, public panic, and economic loss remains significant.

David Relman, a microbiologist and biosecurity expert at Stanford University, highlighted specific risks in the letter. He noted that AI tools enable users to quickly identify where to order sequences that will not be subject to screening. If prompted appropriately, these models can also suggest modifications to orders to evade detection by screening software. This concern is supported by a study from Microsoft researchers, which showed that AI protein design tools were able to generate potentially dangerous gene sequences that slipped past companies’ screening software.

The signatories include executives from gene synthesis firms Twist Bioscience and Ansa Biotechnologies, members of the International Gene Synthesis Consortium formed in 2009 to implement voluntary screening practices. Federal guidelines introduced during the Biden administration already required scientists and companies receiving federal funding to order synthetic gene sequences from providers that screen purchases. A bipartisan bill introduced earlier this year in the Senate would expand this requirement to all gene synthesis providers operating in the US.

Geoff Ralston, former president of Y Combinator, argued that AI labs with biology models should conduct their own screening of users to make it difficult to ask a model to help with imminently dangerous tasks. Relman agreed that regulations around screening procedures are only part of the solution, stating that given the potential for screening failures, other points of control are necessary. He emphasised that AI companies must step up to ensure their technology is used responsibly.