The Chief Executives of The leaders of several artificial intelligence (AI) companies have urged Congress members to pass new legislation that will make it difficult for terrorists to use their technology to produce biological weapons.
Google DeepMind’s Demis Hassabis, OpenAI Sam Altman’s Dario Admodei’s Anthropic, and Microsoft AI Mustafa Suleyman were among the signatures on an a public letter Demanding that laws be passed requiring the screening of customers by companies who sell synthetic DNA or RNA and their orders in order to prevent misuse.
This letter was written by both the left-leaning Foundation for American Innovation and the nonpartisan Institute for Progress. It acknowledges, given the pace at which AI has developed, “there is a real possibility that the knowledge barriers which have historically prevented bad actors from obtaining biological weapons will meaningfully erode.”
Arthur Kornberg, a scientist from the United States in the 1950s was the one to synthesize the DNA. The process has been automated and dozens of commercial synthesizers are used around the globe. “print” They sell genetic sequences for use in scientific research, drug discovery, diagnostics, and other purposes. Most providers limit their sales to researchers, biotech firms, educational institutions and other qualified buyers. But not all of these companies vet the customers they sell or the sequences that are ordered.
Canadian Researchers in 2017 raised alarm When they used 100,000 worth of DNA ordered by mail to reconstitute an extinct horsepox, the scientists were able to re-create the virus. They claimed that this same technique could be applied to smallpox – a similar and deadly virus. Since that time, the cost of gene synthesis has decreased.
Although some biological training is still needed in order to produce a functioning virus, AI and large language models can be used to develop dangerous new pathogens. Although bioterror is rare, it can lead to mass deaths, panic in the public, and financial loss. The AI-designed virus could unintentionally or intentionally spark a worldwide pandemic.
“AI tools enable a user to very quickly identify where to turn to order sequences that will not be subject to screening,” David Relman signed the letter and is a Stanford University expert in microbiology and biosecurity. “If prompted appropriately, they can also tell you how to change the nature of your order, so that even those that are screening may be much less able to detect what it is you’re trying to make.”
Other scientists, national-security experts and executives from Twist Biosciences and Ansa Biotechnologies are also signatories. They are also members of International Gene Synthesis Consortium which formed in 2009. Software is used by many companies to check orders. “sequences of concern” That can cause an organism to become toxic or to have the ability to produce disease.
“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,” James Diggans is vice president for policy and biosecurity of Twist Bioscience. Since years, the company has been in favor of formalizing rules.
Federal guidelines The Biden administration introduced a requirement that required federally funded scientists and businesses to purchase synthetic gene sequences through providers who screen all purchases. When the Biden administration was in power, scientists and companies that received federal funding were required to order synthetic gene sequences from providers who screen purchases. bipartisan bill The Senate introduced a bill earlier this month that would mandate all US gene-synthesis companies to screen customers and orders for harmful pathogens, bad actors and other potential problems.
Screening tools may not be perfect. Microsoft published an article last year that outlined the findings of their research. study AI-based tools that design proteins were found to be able generate dangerous gene sequences, which slipped through companies’ software screening. These models predicted new sequences of proteins that were similar in structure to ones known to be harmful.
Geoff Ralston is a former president of Y Combinator, and he’s a partner in the Safe AI Fund. He believes that AI labs using biology models need to screen their users.
“It should be very difficult, if not impossible, to ask a model to help you do something imminently dangerous,” Ralston also signed the correspondence.
Relman also agrees with the idea that regulating screening procedures only represents a small part of the problem. “Given that the screening may fail in some cases, we must then have other points of control,” He says. “That’s where the AI companies are going to have to step up.”

