Anthropic’s Fable 5 Returns as U.S.-Iran Talks Resume in Doha

The Trump administration has lifted export controls on Anthropic’s Claude Fable 5 and Mythos 5, ending an 18-day suspension that began on June 12 over fears the models could be jailbroken for cyberattacks ^[1][2]. Anthropic said access would begin restoring Wednesday to users globally on Claude platforms, with AWS, Google Cloud, and Microsoft Foundry re-enabling the models “soon” ^[1][3].

The Commerce Department told Anthropic it had “addressed the risks” and agreed to proactively detect security flaws, collaborate on future releases, and alert officials to malicious activity ^[2]. Anthropic says it trained a new safety classifier that blocks the specific jailbreak technique in more than 99% of cases; blocked requests are rerouted to Opus 4.8 ^[1][3]. The episode has become a case study in the ad hoc regulation of frontier AI: OpenAI similarly launched GPT-5.6 Sol last week under a limited preview for government-approved partners, and the administration faces an August deadline to set standardized security benchmarks ^[2][4].

The Fable 5 reversal also comes at a delicate moment for Anthropic, which is preparing an IPO and has been sparring with Washington for months over supply-chain risk designations ^[1]. In a blog post, the company pledged pre-release government access to models with national-security relevance, rapid information sharing on jailbreaks, and a voluntary shared security standard with other labs ^[1].

Meanwhile, world attention is fixed on Doha, where U.S. envoys are meeting Qatari mediators and an Iranian technical delegation to implement the June 17 memorandum of understanding that extended the U.S.-Iran ceasefire by 60 days ^[5]. Tehran denies direct talks are taking place and says it is only discussing the release of roughly $6 billion in frozen assets and other MoU provisions ^[5]. Vice President JD Vance called Iran’s public denials a “Persian negotiation tactic” and warned that any Iranian attacks on shipping in the Strait of Hormuz would draw a U.S. military response ^[5].

The two storylines share a common thread: both Washington and the tech industry are racing to build guardrails—one for nuclear diplomacy, the other for frontier AI—before a fragile truce turns into a wider crisis.