You don’t have to be an AI fan to recognize that sometimes AI can be a useful tool in the search for new knowledge. If any of that new knowledge were turned into a recipe for a bioweapon, I think most people would agree it would suck.
Sam Altman, CEO of OpenAI, and Dario Amodei, CEO of Anthropic I don’t agree with muchbut they agree that that would be bullshit, and signed their names to an open letter saying so. The letter itself calls this “a rare moment of agreement between interested parties who often disagree.”
Other signatories to the letters include Demis Hassabis and Alexandr Wang, heads of AI at Google Deepmind and Meta, respectively, along with other AI entrepreneurs and researchers, plus dozens of scientists and policy experts.
The central idea of the letter actually has nothing to do with AI; is aimed at policymakers and simply calls for legislation to be passed that requires those with the ability to control synthetic nucleic acid to do so. Specifically, it asks that when requests for DNA (and likely RNA, although the letter does not mention this) come in, they be scanned for “sequences of concern” and that “client legitimacy” be verified before the synthesized nucleic acid is mailed. It also calls for data on orders to be recorded and potentially made available to researchers, stating that “knowledge of traceability in itself deters misuse.”
The rapid development of AI only adds urgency. The letter says that because AI is progressing so rapidly, “there is a real possibility that the knowledge barriers that have historically prevented bad actors from obtaining biological weapons will be significantly eroded.”
cabling grades that the letter was organized by two think tanks: the Institute for Progress, which describes itself as nonpartisan, and the Foundation for American Innovation, which is ostensibly right-wing.
OpenAI, for its part, has taken steps lately that seem aimed at associating the company and its leader with responsibility. He threw a policy white paper Tuesday, outlining a plan for research into AI models at the federal level that is stricter than the plan in a President Trump’s recent executive order. On Wednesday, Altman also met with Bernie Sandersthe fiercest critic of AI currently in the US Senate.
