Technology & Science
US Export-Control Order Halts Anthropic’s Fable 5 & Mythos 5 Worldwide
On 12 June 2026, a Commerce Department directive gave Anthropic barely 90 minutes to bar all non-US citizens from its just-launched Fable 5 and Mythos 5 AI models, so the company pulled both models offline for every user worldwide.
Focusing Facts
- Commerce Secretary Howard Lutnick’s letter threatened civil and criminal penalties under export-control statutes unless access for foreign nationals—inside or outside the US—was cut immediately.
- Claude Fable 5 had been released on 9 June 2026 and, days later, overtook OpenAI’s GPT-5.5 Pro on the Epoch Capabilities Index with a score of 161 to 160.
- Earlier, on 4 March 2026, the Pentagon designated Anthropic a "supply-chain risk" after the firm resisted uses involving mass domestic surveillance and autonomous weapons.
Context
Washington last wielded blanket software export bans during the 1990s crypto wars—when PGP was classified as a munition until 1996—yet those rules targeted executable code, not cloud APIs. The new order signals a deeper trend: governments are moving from policing semiconductor supply chains (as with 2022–23 chip controls on China) to policing the models themselves, treating frontier-grade AI as dual-use matériel. It also exposes strategic dependence; EU cybersecurity teams suddenly lost a critical tool, echoing post-OPEC energy anxieties of the 1970s. Over a century horizon, this may mark the point when nation-states re-assert sovereign control over intangible computation, fragmenting the global AI commons much as nuclear technology fractured after 1945—unless, as with strong encryption, technical diffusion ultimately outpaces regulatory walls.
Perspectives
Left-leaning tech media
Left-leaning tech media — Portrays the Trump administration’s export-control order as punitive meddling meant to cripple Anthropic for requesting modest AI regulation. Long-standing distrust of conservative administrations and Big Tech power pushes this press to frame the dispute as political bullying, soft-pedalling the concrete security claims other outlets detail.
Security-focused outlets and hawkish commentators
Security-focused outlets and hawkish commentators — Frame Mythos and Fable as jailbreak-prone cyber weapons whose global availability could aid hostile states, so sweeping export controls are warranted. By foregrounding worst-case scenarios and China-linked fears, they may exaggerate the models’ danger to justify expansive national-security authority over commercial AI.
Business and finance media
Business and finance media — Cast the shutdown chiefly as a material regulatory risk that reshuffles competitive positions and threatens Anthropic’s IPO and backers’ valuations. Market-first framing can sideline ethical or security stakes, echoing investor priorities and giving heavy weight to lobbying access and corporate strategy.
Like what you're reading?