Skip to content
Aerial panorama of the Singapore Strait and the Pasir Panjang container terminal

V Technology Geopolitics & AIAsia-Pacific

SPF argues AI-era biosecurity must track capabilities, not pathogen lists

Singapore Strait and Pasir Panjang Port Terminal (2016)Photo: Bob Tan / Wikimedia Commons · CC BY-SA 4.0 · resized · Source
Institution
Sasakawa Peace Foundation (SPF) — IINA
Author
Kiwako Tanaka
Country · language
Japan · English
Affiliation
Independent foundation (think tank)

Summary

In an article published on SPF's IINA platform on 11 November, Kiwako Tanaka of Toyo Eiwa University proposes changing what biosecurity aims at. In her view, policing lists of dangerous pathogens and toxins is no longer enough. When AI, biological data, expertise, DNA synthesis and laboratories come together, new research capabilities emerge quickly. The author therefore sets the question 'what is becoming possible' alongside 'which pathogen is dangerous'. Her rationale rests on a string of assessments from US and European institutions in 2026.

The main evidence is Anthropic's misuse report of September 2026. According to the report, the company's AI assistant was used for planning, literature review and data analysis in dual-use research on highly pathogenic avian influenza, orthopoxviruses and toxins. Tanaka writes that this does not mean weapons were produced, but it does show experts working faster. She adopts the layered defence idea from RAND's August 2026 report: combining weak signals at several control points, such as AI access, DNA synthesis orders and laboratory services.

For Japan the diagnosis is clear. The Infectious Diseases Control Law governs pathogens, the Foreign Exchange and Foreign Trade Act governs exports, and university rules govern research security; these 3 frameworks answer to different authorities. Before adding new regulation, the author wants an integrated assessment function that brings these signals together. She defines it not as covert intelligence gathering but as combining different sources and presenting conclusions to decision-makers.

Blind spot

The piece relies almost entirely on US and European sources; China's biotechnology and DNA synthesis capacity, the competitive environment closest to Japan, never comes up. It does not discuss which ministry the proposed integrated function would sit under, its budget, or the legal limits of data sharing by companies. The problem of synthesis providers that do not take part in screening is also left out.

Talay assessment

Bottom line

Tanaka's reading turns biosecurity in Japan from a job of maintaining banned lists into one of fusing signals. The 3 existing legal frameworks are spread across different authorities; the proposal is not a new law but an inter-agency assessment unit. The most likely path is that Tokyo folds the issue into its existing economic security and research security agenda and does not set up a separate unit in the near term.

Likely effects

  • AI companiesUncertain1–6 months

    If the trusted-user model spreads, access to advanced biology capabilities will be limited to verified researchers; AI companies effectively become the gatekeepers of access.

  • DNA synthesis supply chainNegative1–6 months

    Synthesis providers become control points through customer identity checks and sequence screening; AI's ability to design sequences not found in nature widens the area where such screening may be blind.

  • TürkiyeUncertain6 months+

    In middle powers with growing biotech capacity, pathogen, export and research rules are often split across separate authorities. The integrated-function question Japan is debating applies to Türkiye too.

Possibilities, ranked

  1. 1
    Added to the existing agenda55%

    Japan adds AI and biology risk as an extra clause in its research security guidelines, and no separate assessment unit is created.

    Watch: Whether Japan's research security guidelines include a clause on AI-assisted biology

  2. 2
    Integrated unit created30%

    An assessment function fusing AI, life sciences and health signals is set up at Cabinet level.

    Watch: A Cabinet Office budget request for a biosecurity assessment function

  3. 3
    Hard ban after an incident15%

    A public misuse case pushes the signals approach into the background and access bans come to the fore.

    Watch: The first public investigation into AI-assisted biological misuse

Probabilities are calibrated judgement based on the sources, not measurement, and are revised as new information arrives. Not investment advice.

Original publication: spf.org · 11 November 2026

This page summarises the institution's view and does not reflect the view of Talay Insight. No direct quotation is used.