The Australian government's AI Biosecurity Office stood up on April 24, 2026 — Day 10 by Wednesday's count. [1] The Australians for AI Safety open letter to Agriculture Minister Julie Collins published April 8 has now drawn over 120 signatories, including parliamentarian Kate Chaney, Dr. Cassidy Nelson of the Centre for Long-Term Resilience, Janet Egan of the Center for a New American Security, Dr. Toby Ord of Oxford, and Associate Professor David Heslop of UNSW. [2] [3] The letter asks Australia to use existing Biosecurity Act 2015 authority to require BICON gene-synthesis import permits to source synthetic nucleic acids only from providers that screen orders for sequences of concern. The Director of Biosecurity has the legal authority to do this without new legislation. [4]
The U.S. counterparty has produced no response. The New York Times bioweapons-transcripts report — the documented finding that frontier AI models can substantively assist non-experts in synthesizing pathogens — published April 29. Day 7 by Wednesday morning. HHS, CDC's biosafety office, the National Security Council, NSF, and NIH have collectively issued no statement. [5] No agency has briefed the Senate Health Committee or House Energy and Commerce. The asymmetry is not a difference of urgency; it is a difference in whether the institutional architecture exists to produce a response.
Tuesday's Day 9 / Day 6 framing read the asymmetry as hardening. Monday's Day 8 / Day 5 framing read the asymmetry as forming. Wednesday's frame is the third clock the prior pieces did not yet carry: the National Science Board, on Day 12 of its disbandment, is the body that under any prior administration would have requested an interagency review of exactly this question. The body does not exist. The architecture for the request is gone.
The NSB's role in this scenario is procedural and consequential. The Board, established under the National Science Foundation Act of 1950, is the federal agency's external advisory body and the empaneled mechanism for triggering coordinated cross-agency reviews on emergent science-policy questions. The Pulitzer Prize Board's 13 former NSF director letter on May 4 named the disbandment specifically: thirteen of the agency's prior leaders signed a public statement demanding the restoration of both an NSF director and the Board itself. [6] The 13-leader letter is the academy-side organized response. It does not have an addressee that can act on it.
What does fill the federal silence in adjacent rooms is professional-society self-governance. The American College of Radiology's annual meeting closed Wednesday with the announcement of the first U.S. medical-imaging AI practice parameter — a clinical-practice document specifying how AI-driven imaging tools should be validated and deployed in radiology. The ACR has roughly 41,000 members and is not a federal agency. Its practice parameters are de facto standards because no one else is publishing them. The pattern that radiology is showing is the pattern that biosecurity will likely have to show next: when the federal layer is silent, the relevant professional body issues a document, and the document becomes the regulation by default.
The Australian model is structurally different. The Albanese government's AI Biosecurity Office uses statutory authority that already exists. The signatories of the AI Safety open letter — Toby Ord, Cassidy Nelson, Janet Egan, the Charles Sturt biosecurity scholars — are mostly the same intellectual community as the U.S. signatories of the 13-leader NSF letter. The infrastructure they are working through is what differs: an active office that can write a permit condition versus a recently-disbanded body whose statutory authority is currently unexercisable.
The Australia letter is concrete. Recommendation 1: a screening condition applied to all BICON import permits for synthetic nucleic acids. Recommendation 2: applicants for higher-risk permits prioritized for the statutory fit-and-proper-person test. Recommendation 3: department review of the regime's adequacy, in consultation with the new AI Safety Institute, civil society, industry, and academia. [4] Members of the International Gene Synthesis Consortium — Twist Bioscience, IDT, GenScript — already screen voluntarily; the change targets the unscreened providers, not the compliant ones. [3] [4]
The U.S. version of this regulatory action would have come, under a functioning system, from a White House Executive Order tasking HHS or CDC's biosafety office, with NSF as a coordinating advisor and the NSB as the body endorsing the convening. The EO does not exist. The HHS or CDC convening does not exist. The NSF advisory function is operating without a director and without a Board. The President's Council of Advisors on Science and Technology has not met on the question. The Senate Health Committee has not held a hearing.
What this resembles, structurally, is the pattern in lost-science thread reporting the paper has tracked since March: the institutions are not failing in real time, they are absent from the time the question requires them to act. Day 7 of U.S. silence on a published documented finding is not abnormal as a planning interval — it is abnormal as an interval after which no institutional response is yet visible. Day 10 in Canberra — versus Day 7 plus Day 12 in Washington — is the productive gap the paper has named since Friday. Wednesday's data is that the gap widened.
The Australia office issues its first formal recommendations to the Director of Biosecurity within days, per the open letter's signatories. The U.S. response, if it arrives, will have to come from a White House that has not used the convening authority on bio-AI risk in this term. The asymmetry is the architecture, not the moment.
-- ANNA WEBER, Berlin