Why it Matters
The U.S. has no single federal law enforcing biosafety and biosecurity beyond the Federal Select Agent Program, according to a Congressional Research Service (CRS) report released Thursday, July 16. That leaves oversight to a patchwork of federal guidance and industry self-governance, even as AI can now design DNA sequences, characterize proteins, and run autonomous lab experiments with minimal human oversight.
In May 2025, the White House issued Executive Order 14292, directing the Office of Science and Technology Policy (OSTP) to revise the 2024 dual-use research policy within 120 days and the synthetic nucleic acid screening framework within 90 days. As of July 2026, CRS has not been able to determine whether OSTP completed either directive.
The Big Picture
The 2024 Framework for Nucleic Acid Synthesis Screening applies only to federally funded work, leaving a gap that industry has partly filled: the International Gene Synthesis Consortium (IGSC) developed a common protocol for its members to screen both gene-synthesis orders and the customers placing them.
AI developers are also self-regulating. OpenAI's May risk framework covers cyber, chemical, biological, radiological, and nuclear threats, and offers up to $50,000 to researchers who find jailbreaks in ChatGPT's biosafety safeguards. Anthropic's June framework proposes government obligations for frontier developers around biological weapons, cyber operations, loss of control, and automated research and development.
Not everyone sees an imminent threat. Nobel laureate chemist Dr. Jennifer Doudna said in a June interview that biology remains complex and that she is not seeing chatbots generate genuinely new ideas, a distinction that matters since AI-generated designs still require physical synthesis and evaluation before they pose any real risk.
Congress has five competing bills but no unified approach. H.R. 3029 would direct the National Institute of Standards and Technology to develop nucleic acid synthesis standards; S. 3741 would address broader biosecurity authorities; S. 4363 would fold engineering biology risks into the biennial biodefense threat assessment; S. 4069 would set AI-ready biological data standards; and S. 3952 would prioritize AI innovation support over risk mitigation.
The Bottom Line
No federal agency holds clear authority over the intersection of AI, biological research, and synthetic biology, and neither existing law nor voluntary industry standards appear built for that convergence. Congress must decide whether to pursue comprehensive legislation or continue with the fragmented bills already on the table, while the underlying question, how much AI can actually contribute to biological weapon design, remains unresolved.
Access the Legis1 platform for comprehensive political news, data, and insights.
Spot something wrong? Report an issue with this article