The United States government has eased some of its restrictions on one of the most advanced artificial intelligence systems in the world, letting select American organizations regain access to Anthropic’s powerful AI model, Claude Mythos 5. The reversal comes weeks after officials suspended the model’s release over national security concerns.
The move marks the latest turn in a sharpening debate over how governments should handle cutting-edge AI technology that promises to reshape entire industries while also carrying real security risks.
Anthropic said Friday that more than 100 trusted U.S. organizations, including major corporations and research institutions, can once again use the advanced system. The change follows weeks of talks between the company and federal officials over safeguards meant to prevent misuse.
The decision also points to a deeper tension facing Washington: keeping America ahead in AI development while making sure powerful models don’t end up in the wrong hands.
Why the Government Halted the Release
Officials first ordered Anthropic to suspend access to its most advanced models, Mythos 5 and Fable 5, on June 12. The concern was that the systems could be used for serious cyberattacks or fall into the hands of foreign adversaries.
Government officials have warned that advanced AI systems can speed up cyber warfare, help identify software vulnerabilities, and strengthen offensive cyber operations. The worry is especially sharp when it comes to countries Washington views as strategic rivals, including China and Russia.
Experts say frontier AI models can now write complex code, spot security weaknesses, and handle tasks that once required entire teams of cybersecurity specialists. That shift has pushed the Trump administration toward a much tighter approach to monitoring and regulating how American AI companies release their most powerful systems.
Anthropic’s Statement
Anthropic confirmed Friday that the government approved a limited return of Mythos 5. The company said the government notified it that Mythos 5, described as its strongest cybersecurity model, can be redeployed to a group of U.S. organizations that operate and defend critical infrastructure.
Anthropic said it’s moving quickly to restore access for approved organizations and continuing to work with officials on expanding availability down the road. The company also said it hopes restrictions on Fable 5, another advanced system still unavailable to the public, can eventually be loosened.
Trusted Organizations Gain Access

People familiar with the new directive say more than 100 companies and institutions can now use Mythos 5. Many belong to Project Glasswing, an Anthropic initiative that includes major tech firms, research organizations, and Fortune 500 companies.
The policy also lets non-U.S. citizens use the model, as long as they work for an approved organization or for Anthropic directly. Companies and institutions left off the government’s approved list still face licensing restrictions.
That selectiveness has drawn criticism from civil liberties advocates and some tech industry leaders.
Critics Question Government’s Role
The government’s authority to decide which companies get access has raised questions about transparency and fairness.
John Coleman, legislative counsel at the Foundation for Individual Rights and Expression, said the process lacks clear standards. “No one knows how these companies are picked and why everyone else is excluded,” he said. “This is putting too much power in the hands of the government. There’s little transparency and it raises serious questions about the rule of law.”
The criticism feeds into a wider worry that federal authorities could end up with outsized control over who benefits from emerging AI technology.
OpenAI Raises Similar Concerns
Anthropic isn’t the only company facing this kind of scrutiny. OpenAI announced Friday that it’s delaying the wider release of its newest model, GPT-5.6, at the government’s request. Access will start with a small group of vetted partners.
OpenAI CEO Sam Altman raised his own concerns about the arrangement. “Extensive safety testing is not a bad idea,” he wrote on social media. “I just don’t like the idea of the government picking the customers.” His comment reflects a broader unease across the tech sector about how far government involvement in AI deployment should go.
National Security and Cyber Risks
Security experts say the government’s caution has real grounding. Advanced AI models have shown growing skill in coding, network analysis, and cybersecurity research. In the wrong hands, that skill could help identify weaknesses in critical infrastructure, automate cyberattacks, build sophisticated malware, or speed up espionage. Financial systems, power grids, telecom networks, and transportation systems are all potential targets.
Commerce Secretary Howard Lutnick wrote in a letter to Anthropic that “significant progress” had been made addressing these risks. Neither the government nor Anthropic has said publicly what specific safeguards are now in place.
The Fable 5 Question
While Mythos 5 is returning to limited use, Fable 5’s future is still unclear. Both models run on the same underlying technology, but Fable 5 was built for wider public release.
Officials reportedly grew concerned after learning that safeguards meant to prevent misuse of the system could be bypassed through jailbreak techniques. The administration is said to be weighing a future release of Fable 5, though no timeline has been set.
Anthropic’s Complicated Relationship with Washington
Anthropic’s dealings with the federal government have involved both cooperation and friction. The company has refused to let its AI systems be used for domestic surveillance or fully autonomous weapons, a stance that has put it at odds with parts of the government. At one point, it landed on a national security watch list over disagreements about military uses of its technology.
Even so, Anthropic has kept working with federal officials on safety measures and frameworks for responsible AI deployment.
A New Era of Government Oversight
The restrictions on Anthropic and OpenAI follow an executive order President Trump signed creating a voluntary review framework for advanced AI models. Under it, companies can submit “covered frontier models” for government review before releasing them to trusted partners.
The administration says this step is necessary to keep dangerous technology from being weaponized. Some experts counter that overly tight restrictions could slow American innovation and hand an advantage to foreign competitors.
Kate Koren, an analyst at the Center for Strategic and International Studies and a former Commerce Department official, called the current policy a temporary fix. “The longer there isn’t a system in place that allows U.S. companies to widely release new models, the more likely it is that China will be able to catch up,” she said.
The Global AI Race Intensifies
The fight over Mythos 5 reflects a bigger picture: artificial intelligence is now treated as a strategic asset on par with nuclear technology or advanced semiconductor manufacturing. Governments around the world are racing to lead in AI development while trying to keep the technology from becoming a new source of risk.
The partial return of Mythos 5 access shows Washington trying to find that balance, protecting national security without choking off innovation. Whether that balance holds as AI systems grow more powerful remains one of the central questions of this moment.
George Mensah is a journalist covering global politics, international conflicts and economic developments for clicxpost. He specializes in breaking news analysis and geopolitical reporting.















