News · 4 November 2026
The AI export-control article is out in Frontiers in Political Science
“Enforcing AI export controls in the digital age: trust infrastructure after the 2026 Fable 5 recall” is published in Frontiers in Political Science, volume 8, article 1908543, doi:10.3389/fpos.2026.1908543, as a Policy and Practice Reviews article in the Politics of Technology section. It appeared online on 4 November 2026 and is open access under CC BY 4.0. The journal received the manuscript on 14 June 2026 and accepted it on 17 August 2026.
In June 2026 a directive scoped to foreign nationals reached two frontier AI systems. By the provider's own account, it had no reliable way to verify nationality in real time, so it suspended both systems for every user until the controls were lifted on 30 June. The article reads this as a case about mechanism: export controls for frontier AI depend on deciding, at the moment of use, who may receive a capability and whether that grant is still valid, and today's digital services do not expose that decision.
Its theoretical contribution is a third condition on chokepoint coercion, beside jurisdiction and domestic institutions: executable granularity, a node's capacity to deny at the granularity the order names. Precise, requester-scoped enforcement then needs 2 things at the point of decision: purpose-bound evidence of the requester's legally relevant attributes, and a versioned authorization mechanism that maps those attributes to entitlements and leaves an auditable decision. Public status registries and recognition across jurisdictions add consistency, portability and oversight, but they are not preconditions of that minimum.
The 2017 Estonian ROCA eID-card recall serves as a bounded comparator, with its preconditions stated rather than assumed. The closing design sketch names the surveillance, discrimination and due-process hazards such a mechanism creates.
The evidence boundary is explicit: the episode is documented and provider-reported, not independently verified, and the article treats the suspension and restoration sequence as within-case evidence, not as an experiment.