AI Chiefs Back Controls as Biological Arms Threat Grows

Top artificial-intelligence executives and biosecurity specialists are urging Congress to require screening of synthetic DNA and RNA orders, warning that AI could lower the technical barriers to developing biological weapons.

The open letter was signed by OpenAI Chief Executive Sam Altman, Anthropic Chief Executive Dario Amodei, Google DeepMind Chief Executive Demis Hassabis, Microsoft AI Chief Executive Mustafa Suleyman, Meta AI chief Alexandr Wang, and other technology, biotechnology and national-security figures. The signatories called for mandatory screening of synthetic nucleic acid orders, customer verification and recordkeeping by providers and manufacturers of synthesis equipment. 

The target is a narrow but important point in the biosecurity supply chain: companies that “print” or sell synthetic genetic material. The letter says providers should check requested sequences for potential concern, verify the legitimacy of customers before shipping, and keep records so that dangerous orders can be traced even if initial screening fails. 

The appeal builds on the Biosecurity Modernization and Innovation Act, introduced Feb. 4 by Sens. Tom Cotton, R-Ark., and Amy Klobuchar, D-Minn. That bill would direct the Commerce secretary to require gene-synthesis providers to screen orders and customers for “bad actors or dangerous pathogens,” and would establish a NIST biotechnology governance sandbox for testing biosecurity tools. 

The concern is not that AI alone can create a pathogen, but that it may make dangerous biological design, procurement and troubleshooting easier for less sophisticated actors.

The letter states that AI systems are improving quickly and that “knowledge barriers” that have historically limited access to biological-weapons capabilities may erode. 

Existing U.S. policy is partial. HHS issued 2023 screening guidance encouraging sequence screening, customer verification, follow-up review for sequences of concern, and recordkeeping, but the framework is largely guidance-based and tied most directly to federally funded research rather than a universal statutory mandate.