2026-07-01 Changes confirmed high confidence

Anthropic Restores Fable 5 Globally After US Lifts First-Ever Export Controls on a Deployed Model

The Commerce Department's June 30 reversal ended an 18-day suspension and set the template for government gating of frontier AI — with Anthropic adding a new cyber classifier as a condition of return.

Anthropic restored global access to Claude Fable 5 on July 1, 2026, after the US Commerce Department lifted export controls on June 30 that had suspended the model worldwide since June 12, according to Anthropic's redeployment announcement and coverage by CNBC, Bloomberg, TechCrunch and others catalogued by LetsDataScience.

Context

The June 12 suspension — directed by the Bureau of Industry and Security under ECRA authority — was the first time export controls were used against a model already in deployment. It followed reporting that a jailbreak technique allowed Fable 5 to assist with certain cyber-exploitation tasks. The suspension lasted from June 12 until the June 30 lifting, with restoration effective July 1; day-count framings in press coverage varied (18 to 20 days) depending on endpoints.

What changed

• Global restoration, no nationality verification. Fable 5 returned across Claude.ai, Claude Platform/API, Claude Code and Cowork from July 1; availability on AWS, Google Cloud and Microsoft Foundry was re-enabled on separate timelines.
• New safeguard. Anthropic deployed a cybersecurity classifier that blocks the reported jailbreak technique in more than 99% of cases — a figure Anthropic relayed that has not been independently audited. Blocked cyber queries reroute to Claude Opus 4.8.
• Goodwill pricing. Fable 5 usage counted toward up to 50% of weekly paid-plan limits, initially through July 7, later extended to July 19. Terms then changed again on July 20 (see update note).
• Mythos 5 stayed restricted. The more capable Mythos 5 remained limited to Project Glasswing participants. Reported cohort sizes conflict — roughly 12 vetted companies in some accounts versus about 100 US organizations approved June 26 per CNBC — likely reflecting a core consortium versus expanded approvals.

Why it matters

The suspension-and-reversal cycle is the access-control story of 2026: it proved that a government can switch a deployed frontier model off, that an 18-day outage stress-tests enterprises built on a single model, and that remediation commitments — Anthropic reportedly agreed to risk-detection, information-sharing and standards-cooperation measures — can switch it back on. Commerce's reversal letter from Secretary Lutnick to Anthropic's Tom Brown struck a cooperative tone, per July 2 reporting.

Details

Anthropic's counter-argument — that the same jailbreak reproduced on GPT-5.5, Opus 4.8 and Kimi K2.7 — is covered in a companion post. CNBC reported that executives warned the ban handed Chinese labs catch-up time during the suspension window.

Limitations and caveats

The >99% classifier-block figure is Anthropic-relayed; direct verification of the classifier's performance was not possible at publication. Glasswing cohort numbers remain unreconciled. Post-restoration terms shifted twice in three weeks, so pricing details should be re-checked before relying on them.

Sources

• Aivy — Anthropic's "Redeploying Fable 5" announcement (quoted) (official text via secondary)
• LetsDataScience — Fable 5/Mythos 5 export ban lifted: press coverage catalogue (aggregator)
• AIToolsRecap — July 2 AI news: controls lifted, Lutnick letter (aggregator)

*Update note: This post was last reviewed on 2026-07-22. Anthropic announced July 18 that, effective July 20, Fable 5 became a standing entitlement of Max and Team Premium plans at up to 50% of weekly usage limits; Pro and Team Standard moved to metered usage credits with a one-time $100 credit, and Claude Code's +50% weekly-limit boost was extended to August 19.*

Drafted with AI assistance from source briefs; reviewed for citation completeness and label accuracy.