AI leaders call for tougher protections against AI-aided bioweapons

Leaders from OpenAI, Anthropic, Microsoft, Meta, and Google DeepMind are urging Congress to mandate screening of synthetic DNA and RNA orders, warning that AI tools are lowering barriers to designing biological weapons [1]. The open letter, signed by Dario Amodei, Sam Altman, Mustafa Suleyman, Alexandr Wang, and Demis Hassabis, calls for closing what they describe as an alarming biosecurity gap [1]. Companies selling synthetic genetic material are not currently required to screen purchases for sequences that could be used to make dangerous pathogens, a voluntary practice the signatories want made mandatory [1]. The letter was also signed by scientists, national security experts, and executives from biotech firms including Twist Bioscience and Ansa Biotechnologies, and was organized by the Foundation for American Innovation and the Institute for Progress [1]. “Given the pace at which the underlying technology is changing, we believe the need is urgent,” the letter states [1]. The signatories argue that AI models could make it easier to design potentially dangerous sequences and order them from manufacturers, work that previously required specialized expertise and access to sophisticated labs [1]. Detailed records should be kept on any orders to track threats that evade initial screening, the letter adds [1]. A recent analysis of 1,178 safety and reliability papers from generative AI research between January 2020 and March 2025 found that corporate AI research increasingly concentrates on pre-deployment areas such as model alignment and testing, while attention to deployment-stage issues has waned [5]. The study identified significant research gaps in high-risk domains including healthcare, finance, and misinformation, and recommended expanding external researcher access to deployment data [5]. The biosecurity push arrives as major tech companies are projected to spend an estimated $650 billion on AI data centers in 2026, according to industry tracking [11]. These specialized facilities, optimized for the parallel processing demands of AI workloads, have become a focal point of the broader AI infrastructure buildout [11]. The letter’s signatories frame the screening requirement as a rare moment of agreement across stakeholders that are often at odds [1].