Labs

U.S. lifts export
controls on Claude
Fable 5 and
Mythos 5, ending
18-day shutdown

The Commerce Department cleared Anthropic's two most powerful models on June 30 after a three-week standoff triggered by an Amazon-documented jailbreak. Fable 5 returns globally July 1; Mythos 5 stays limited to a vetted U.S. cohort.

By Casey Vanderhoof Frontier labs correspondent · July 2, 2026

The U.S. Department of Commerce lifted export controls on Claude Fable 5 and Mythos 5 on June 30, ending an 18-day worldwide shutdown of Anthropic’s two most advanced models. Fable 5 begins a staged return across Claude.ai, Claude Code, and Claude Cowork on July 1, with AWS, Google Cloud, and Microsoft Foundry re-enabled as quickly as their pipelines allow. Through July 7, Pro, Max, Team, and select Enterprise subscribers receive Fable 5 covering up to 50% of weekly usage limits before credits apply. Blocked requests will route to Claude Opus 4.8, per Quartz.

Mythos 5 isn’t fully back. A partial lift on June 26 restored access for roughly 100 vetted organizations, Nextgov reported, and further distribution runs through Project Glasswing, Anthropic’s cyber-defense program. The NSA, which had been using the latest Mythos build, was among the users caught by the original order.

That order, issued June 12, followed a report from Amazon researchers describing a method of prompting Fable 5 to identify software vulnerabilities and produce exploit code. Anthropic disputed the framing. Its own testing, disclosed Tuesday, found that less capable models including Claude Opus 4.8, GPT-5.5, and Kimi K2.7 could identify the same vulnerabilities, and that every model tested, GPT-5.4 and Claude Haiku 4.5 included, could reproduce the exploit demonstration. The company effectively argued the jailbreak was an industry condition, not a Fable-specific defect.

The government wanted concessions anyway. Per TNW, Anthropic agreed to proactively detect and address security risks, help develop standards for future models, and report malicious activity to the government. A new safety classifier neutralizes the Amazon-documented bypass in more than 99% of attempts in testing, according to Quartz, and Anthropic is opening a HackerOne jailbreak program alongside expanded early access for the Commerce Department’s Center for AI Standards and Innovation.

Both sides moved to declare victory in public. “Over the past two weeks, we have worked closely with Anthropic to analyze and approve Fable 5 to ensure alignment across the US Government and strengthen America’s leadership in AI,” Commerce Secretary Howard Lutnick posted on X. White House chief of staff Susie Wiles added: “Our shared priority remains: get the best tech deployed as quickly and safely as possible.”

The commercial backdrop makes the choreography legible. Anthropic has filed a confidential IPO prospectus disclosing a $47 billion revenue run rate and a $965 billion valuation, Quartz reported. And per the same account, co-founder Tom Brown, not CEO Dario Amodei, took the primary negotiating role with Commerce, a dynamic reflected in Lutnick’s correspondence being addressed to Brown. The lesson from the last 18 days, Anthropic conceded, is that the industry lacks a consistent approach to evaluating and responding to jailbreaks. The resolution just delivered is bespoke, bilateral, and roughly the shape of every AI export dispute to come.

Sources