CSIS Report Flags AI Bioterrorism Risk as White House Biosecurity Posts Sit Empty

The federal government's capacity to screen AI-assisted bioterrorism threats is eroding at the precise moment the threat surface is expanding, according to a report reviewed by this reporter and published by the Center for Strategic and International Studies.

The CSIS analysis, authored by Georgia Adamson and Gregory C. Allen of the Wadhwani AI Center, identifies two converging risks: large language models that can lower the informational barriers to bioweapon development, and biological design tools - purpose-built AI systems capable of generating novel protein or nucleic acid sequences - that may be approaching a threshold where they provide actionable uplift to actors with limited technical training. The report stops short of claiming that capability has been crossed; it assesses, at moderate confidence, that commercial AI models are "on the cusp" of meaningfully helping novices develop and acquire bioweapons by providing step-by-step guidance.

That dual-use ambiguity is the honest center of the story. The same biological design tools that could accelerate drug discovery or pandemic countermeasure development could be repurposed. The CSIS report does not resolve that tension falsely; it flags it and argues for layered defenses across multiple steps of what it calls the "bioweaponization pathway."

What makes the report land harder than the usual think-tank output is its budget context. The FY 2026 White House budget proposal, as described in the CSIS analysis, threatens cuts of $18 billion to the National Institutes of Health, $3.6 billion to the Centers for Disease Control and Prevention, and $240 million to the Administration for Strategic Preparedness and Response. The proposal also includes a $325 million reduction to the National Institute of Standards and Technology - the same agency the administration's own AI Action Plan designates as a primary contact for evaluating frontier AI models for biological and national security risks.

The report's framing of that contradiction is blunt: the administration is simultaneously expanding NIST's mandate and proposing to gut its budget. NIST houses the Center for AI Standards and Innovation, or CAISI, which under the AI Action Plan is assigned responsibility for leading unclassified evaluations of AI capabilities in areas including biosecurity and chemical weapons. Whether CAISI can execute that mission under the proposed funding levels is, at this point, an open question - one the report flags without resolving.

The personnel picture compounds the structural one. The CSIS report notes that proposed budget cuts coincide with "dwindling staff across key federal biosecurity-related agencies" and a leaderless White House Office of Pandemic Preparedness and Response Policy. That vacancy traces back to the July 2025 resignation of Dr. Gerald Parker, who had been serving as NSC senior director for biosecurity and pandemic response. According to reporting by STAT News, Parker resigned after roughly six months and was never formally appointed director of OPPR itself. By the time Parker left, according to reporting in The Hill, OPPR had gone from 20 staff under the previous administration to zero full-time employees. The administration has made no public move to fill the position.

Congress established OPPR under the PREVENT Pandemics Act with bipartisan support to serve as the coordinating authority for pandemic preparedness across the executive branch. The office's current status - no director, no staff - is not a gray area. It is a measurable absence.

The CSIS report makes three recommendations: fund NIST and CAISI at levels proportionate to their expanded mandates; have CAISI lead formal evaluations of frontier biological design tools with support from the interagency TRAINS Taskforce and international AI Safety Institutes; and direct OSTP to develop an AI-enabled DNA synthesis screening system capable of detecting novel, AI-generated sequences that fall outside existing static lists of regulated agents and toxins. The third recommendation addresses a known gap: current list-based screening can only flag known threats, while biological design tools are by definition capable of generating sequences that do not yet appear on any list.

None of these recommendations are technically complicated. All three involve agencies that already exist and have the relevant expertise. The constraint is not technical; it is political and fiscal. Whether the same administration that proposed the cuts will reverse course on them is an assessment this reporter will leave at low confidence.

Sources cited:
- Center for Strategic and International Studies - "Opportunities to Strengthen U.S. Biosecurity from AI-Enabled Bioterrorism: What Policymakers Should Know" (https://www.csis.org/analysis/opportunities-strengthen-us-biosecurity-ai-enabled-bioterrorism-what-policymakers-should)
- STAT News - "Top White House pandemic preparedness official resigns, officials say, in sign of broader disarray" (https://www.statnews.com/2025/07/30/white-house-pandemic-preparedness-office-leaderless-unprepared/)
- The Hill - "White House pandemic team reduced to one part-time staff" (https://thehill.com/opinion/healthcare/5473546-white-house-pandemic-preparedness-collapse/)
- NIST - Center for AI Standards and Innovation (CAISI) (https://www.nist.gov/caisi)