Microsoft has backed stronger screening rules for nucleic acid synthesis in response to biosecurity risks tied to artificial intelligence, casting the issue as a national security priority as AI tools spread through biological research.
Advances in general-purpose AI models, specialist biological design software, laboratory automation and agentic systems are beginning to converge in ways that can accelerate life sciences research while lowering barriers to harmful misuse. Those developments can aid work in fields such as drug discovery and materials science, but they can also open new paths to redesign toxins and pathogens.
At the centre of Microsoft's argument is nucleic acid synthesis screening, which synthetic DNA providers use to check orders and verify customers before turning biological designs into physical material. The company described this checkpoint as one of the clearest near-term controls because it targets access to sensitive material rather than regulating scientific ideas or legitimate research.
Microsoft pointed to uneven standards across DNA synthesis providers, with screening practices still largely voluntary. As AI design systems grow more sophisticated and easier to use, it warned, those gaps between providers will become more significant.
The company cited research showing that specialist AI tools for protein design can be used to re-engineer toxins in ways that may preserve harmful function while evading some existing synthesis safeguards. In its view, that exposed weaknesses in screening systems built for an earlier stage of technological development.
Microsoft also highlighted its Paraphrase Project, which tested current screening systems against AI-designed biological sequences. It said the work identified where safeguards could fail and how they could be improved, using an approach similar to cybersecurity practices such as red teaming, responsible disclosure and rapid fixes.
Policy pressure
The debate has moved further into US policy circles as lawmakers and administrations of both parties have sharpened their focus on biosecurity oversight. Microsoft said support for tighter controls has carried across successive administrations and become part of a broader discussion about governing frontier technologies.
The company pointed to an executive order from the Trump administration on improving the safety and security of biological research, which emphasised nucleic acid synthesis screening and broader biosecurity oversight. It also referred to earlier work by the White House Office of Science and Technology Policy that set out a federal framework centred on comprehensive screening, customer verification and technical standards developed with industry.
Congress is also considering legislative action. Microsoft highlighted the Biosecurity Modernization and Innovation Act, introduced by Senators Tom Cotton and Amy Klobuchar, which would create mandatory screening requirements, conformity assessments and enforcement measures, while also providing technical assistance and a biotechnology governance sandbox for exploratory work.
The proposed legislation would also direct the Office of Science and Technology Policy to assess existing biosecurity authorities and develop a plan to consolidate oversight. Microsoft said the measures point to an emerging consensus in Washington that biotechnology safeguards need updating as AI systems become more deeply embedded in biological research and development.
Converging tools
A central part of Microsoft's case is that the risks do not arise from any one technology in isolation. Instead, it described a growing capability stack in which more powerful generalist AI models make specialist biological tools easier to use, automated laboratories expand access to complex workflows, and agentic software links design, analysis and synthesis into a more continuous process.
That interaction matters, Microsoft argued, because it could allow less experienced actors to move more easily from a digital biological design to real-world synthesis, whether through commercial nucleic acid synthesis services or increasingly automated laboratory systems. In its view, that changes the practical risk landscape even if each individual component has legitimate uses.
Governance therefore needs to cover multiple control points rather than focusing only on frontier AI models, the company said. Screening at the synthesis stage offers a pragmatic route because providers sit at the point where abstract designs can become tangible biological material.
Microsoft framed the issue as one of modernising existing safeguards rather than slowing scientific progress. Biosecurity concerns, it argued, should lead to stronger defences, not reduced innovation, and early technical and policy safeguards are needed before misuse outpaces oversight.
"Strengthening nucleic acid synthesis screening is a pragmatic and targeted response," Microsoft said.
"It does not regulate ideas or restrict legitimate research.
"Instead, it focuses on responsible access to sensitive capabilities, reinforcing a line of defense that already exists but must now be modernized."
The company also said the US has an opportunity to combine scientific leadership with tighter stewardship of fast-moving technologies. "The United States has an opportunity to continue to lead by pairing innovation with responsible stewardship," Microsoft said.