The path to a global relaunch
Last week, authorities allowed a small group of American cybersecurity firms to access Mythos 5, a limited opening that appears to have served as a test run for a broader lifting of controls. In a letter seen by Reuters, Commerce Secretary Howard Lutnick said the restrictions were lifted after Anthropic agreed to specific conditions. The full terms of that agreement have not been made public. Lutnick’s letter laid out a framework that satisfied the department’s national security concerns, according to Reuters, though what that framework looks like in practice remains unclear. Anthropic had argued that restricting access for its own employees–many of whom are foreign nationals–made it nearly impossible to maintain or improve the models. The June 12 order came without warning, catching the company mid-deployment. Teams that had been preparing for a global launch suddenly found themselves locked out of their own systems. Foreign-born researchers, who make up a significant portion of the company’s technical staff, watched from the sidelines as American colleagues scrambled to keep the models stable. By nightfall Tuesday, the mood inside Anthropic’s offices had shifted from frustration to relief as engineers prepared for a global relaunch.
Broader context and company background
The decision comes amid a broader U.S. government push to control advanced AI exports. OpenAI was also asked to limit the rollout of its latest GPT-5.6 model to vetted partners only, signaling that Washington remains deeply wary of how these systems might be used by foreign actors. The Trump administration, with President Donald Trump in office and Vice President JD Vance, has signaled it will take a hands-on approach to advanced AI governance. Secretary of State Marco Rubio’s department has been involved in parallel discussions about the geopolitical implications of these technologies. For Anthropic, the immediate future is about getting Claude Fable 5 back into the hands of users worldwide. The company’s post on X made clear that Wednesday’s relaunch is global, with no more restrictions on foreign nationals and no more suspended access.
Anthropic was founded in 2021 by a group of former OpenAI employees, including siblings Daniela and Dario Amodei, and operates as a public benefit corporation with a stated mission of promoting AI safety. Its flagship product, Claude, is a series of large language models designed to generate human-like text and perform complex reasoning tasks. The company finished training the first version of Claude in 2022 but delayed its public release, citing a need for further internal safety testing and a desire to avoid accelerating a potentially hazardous race to build ever-more-powerful AI systems. Claude was eventually released in 2023. By mid-2025, the company had announced Claude 4, introduced new API capabilities, and launched a web search tool that allowed Claude to access real-time information from the internet. That same year, Anthropic struck a two-year contract with the U.S. Department of Defense, integrating its language models with the Pentagon’s classified intelligence networks. The company stipulated that the military must not use its technology for mass surveillance of Americans or for creating autonomous weapons. In September 2025, Anthropic announced it would stop selling its products to entities majority-owned by Chinese, Russian, Iranian, or North Korean groups, citing national security concerns. The company later signed a cloud partnership with Google, giving it access to a large number of custom processing units, and secured major investments from Nvidia and Microsoft.
The recent back-and-forth over export controls on Claude Fable 5 and Mythos 5 reflects a broader pattern in which Washington is wrestling with how to balance technological leadership, national security, and the free flow of AI research. The tension matters because the same models that promise breakthroughs in medicine, coding, and translation also carry risks that governments are still learning how to manage. The outcome of these skirmishes between companies and regulators will shape who gets access to cutting-edge AI and under what conditions. For now, Anthropic’s engineers are back at work, and Claude Fable 5 is heading out the door.
{“@context”: “https://schema.org”, “@type”: “NewsArticle”, “headline”: “U.S. Eases Controls on AI Model Access in Limited Test”, “description”: “The path to a global relaunch Last week, authorities allowed a small group of American cybersecurity firms to access Mythos 5, a limited opening that…”, “datePublished”: “2026-07-01T09:30:00+00:00”, “author”: {“@type”: “Person”, “name”: “Merry Gel Sigui”}, “publisher”: {“@type”: “Organization”, “name”: “InfoPulse Today”, “url”: “https://newsfeed.infopulsetoday.com/”}}
