US government allows Anthropic limited release of AI model that sparked cybersecurity concerns

US Government Lifts Export Ban on Anthropic AI Model Amid Cybersecurity Concerns

Regulatory Shift Enables Controlled AI Deployment

US government allows Anthropic limited release – The U.S. government allows Anthropic limited release of its advanced Mythos AI model, marking a pivotal step in balancing innovation with national security. On June 26, 2026, federal authorities announced the partial lifting of export restrictions, enabling the company to distribute the model to select domestic partners. This decision follows months of scrutiny over the potential cybersecurity risks posed by Mythos, particularly its capacity to identify and exploit vulnerabilities. The revised framework prioritizes controlled access, ensuring the AI’s most powerful features are available to trusted entities while minimizing exposure to foreign adversaries.

Anthropic’s Response and Safeguard Implementation

Anthropic expressed relief over the decision, highlighting its efforts to address security concerns. In a statement, the company noted that “appropriate measures have been taken to enable trusted partners to access the Claude Mythos 5 model,” aligning with the U.S. government’s approval of a limited release. While the export ban on Fable, a less capable variant, remains in place, the move signals progress in resolving disputes over the model’s deployment. The company is now finalizing a list of approved organizations, aiming to restore access swiftly while maintaining collaboration with regulatory bodies.

“We are pleased to see this progress and continue to work with the government to expand access to Mythos 5 and make Fable 5 available for general use again,” Anthropic said in its release. The statement underscored the company’s commitment to security, emphasizing that its safeguards are now deemed sufficient to mitigate risks. However, the partial release also reflects ongoing negotiations, with officials cautious about broader distribution until additional measures are confirmed.

Analysts note that the revised policy underscores the administration’s growing focus on AI’s strategic role in both defense and economic growth. By allowing Anthropic limited release, the government aims to harness the model’s capabilities without compromising security. This phased approach addresses fears that Mythos could be weaponized by malicious actors, as its ability to rapidly analyze and exploit vulnerabilities has raised alarms. The decision follows a temporary suspension of access to both Mythos and Fable earlier in June, which had restricted foreign users from employing the technology.

Industry experts argue that the regulatory shift reflects a nuanced strategy in AI governance. While the U.S. government allows Anthropic limited release, it maintains oversight to ensure the model’s deployment aligns with national interests. The limited rollout targets critical infrastructure providers and cybersecurity specialists, who are deemed less likely to misuse the system. This method allows for continued innovation while preserving a buffer against potential threats, as officials weigh the benefits of AI advancements against their risks.

Earlier in 2026, Anthropic faced designation as a “supply chain risk,” effectively blacklisting its operations. The company contested this label, leading to legal battles that ultimately yielded partial victories. Now, with the export restrictions eased, Anthropic is working closely with federal agencies to refine its compliance protocols. The limited release not only addresses immediate cybersecurity concerns but also sets a precedent for how the U.S. might regulate other AI models, including those from OpenAI, which has similarly been asked to limit capabilities for security reasons.

As discussions over the limited release continue, the administration is exploring further safeguards to protect against misuse. Sources indicate that the policy adjustment comes after extensive deliberation, with officials prioritizing security without stifling technological progress. The U.S. government allows Anthropic limited release, positioning the model as a key asset in national defense and economic competitiveness. This approach exemplifies the delicate balance between fostering innovation and ensuring that AI technologies remain under secure control.