Trump Admin releases Anthropic Mythos to be used by more than 100 US companies, agencies

In a pivotal development for the artificial intelligence landscape, the Trump administration has eased its stringent export controls on Anthropic's highly capable Claude Mythos 5 AI model. This decision now permits more than 100 US companies and government agencies to utilize the advanced AI, including access for their non-American employees. The move follows a period of intense negotiations and marks a significant moment in the ongoing dialogue between frontier AI developers and national security interests.

A Breakthrough for Claude Mythos 5 Amidst Tense AI Policy

Anthropic's Claude Mythos 5, a cutting-edge large language model specifically engineered for complex tasks in cybersecurity and scientific research, has been at the center of a high-stakes policy debate. Its initial unveiling in June 2026, alongside a more generally available version called Fable 5, was met with immediate regulatory scrutiny from the US government.

Just two weeks prior to this recent announcement, the Trump administration had imposed export restrictions on both Mythos 5 and Fable 5. These controls stemmed from serious national security concerns, including reports of a group potentially linked to China gaining unauthorized access to Mythos, and fears that the models could be "jailbroken" or exploited for malicious purposes. In response, Anthropic temporarily took both models offline worldwide and limited access to foreign nationals, creating a ripple effect across its global partners.

Intensive Negotiations Pave the Way for Access

The lifting of these restrictions comes after a period of "intense, daily talks" between Anthropic and the Commerce Department. Commerce Secretary Howard Lutnick confirmed the decision in a letter to Anthropic's chief compute officer, Tom Brown, stating that "appropriate safeguards are in place" to allow "certain trusted partners" to access the Claude Mythos 5 model. Anthropic, for its part, has committed to working closely with the US government on "protocols and standards and releases" for its models, indicating a collaborative approach to managing the inherent risks of powerful AI.

This resolution represents a crucial de-escalation in the confrontation between the Trump administration and one of the world's leading private AI companies. The approval extends access to a select list of over 100 US institutions, encompassing both major corporations and government agencies. Importantly, the new arrangement also clarifies that a license will no longer be required for exporting, re-exporting, or in-country transfers of Mythos 5 to these identified entities and their foreign national employees, as well as to Anthropic's own foreign national employees.

Mythos 5: A Tool for Critical Applications

Claude Mythos 5 is distinguished by its advanced capabilities, particularly in identifying vulnerabilities in software and generating scientific hypotheses. It has shown significant gains in cybersecurity, biology, and healthcare benchmarks. Its precursor, Claude Mythos Preview, was initially launched through "Project Glasswing" in April 2026, a collaborative initiative involving tech giants like Amazon Web Services, Apple, Google, Microsoft, and NVIDIA, aimed at securing critical software infrastructure.

The model's ability to consistently produce novel and compelling scientific hypotheses, outperforming previous models in molecular biology, highlights its potential to accelerate research and development in critical fields. However, its immense power also brings inherent risks, which Anthropic has acknowledged, stating that Mythos-class models have reached a threshold where they present "significant risks" if misused.

Fable 5 Remains Under Restriction

While Mythos 5 now has expanded access, its counterpart, Claude Fable 5, remains under government restriction for general use. Fable 5 is described as a weaker version of Mythos, initially intended for broader public availability and briefly considered the most powerful AI model widely accessible to consumers. The current letter from the Commerce Department does not address Fable 5, though sources close to the discussions suggest talks are ongoing for its eventual release.

The distinction between the two models lies in their safeguards. Fable 5 incorporates robust safeguards for cybersecurity and biology, automatically routing queries in these sensitive domains to Anthropic's less powerful Opus 4.8 model, effectively making it safer for general consumption. Mythos 5, with fewer inherent guardrails for these specific high-risk areas, is reserved for vetted partners operating within controlled environments.

Broader Implications for AI Governance and Enterprise Adoption

This episode underscores the rapidly evolving role of government in regulating frontier AI technologies. The Trump administration's intervention, initially imposing export controls and now lifting them under specific conditions, signals a growing emphasis on national security and responsible deployment of powerful AI. It reflects a global trend where governments are increasingly involved in vetting companies and setting parameters for access to advanced AI, highlighting a shift from a purely hands-off approach.

The situation also brings to light Anthropic's strategic focus on the enterprise and government sectors. The company has actively pursued partnerships with federal agencies, even offering its Claude for Enterprise and Claude for Government models for a nominal fee to US government branches in 2025. This enterprise-first approach is driven by the need for robust security, data governance, and compliance, which Anthropic has built into its offerings.

Ongoing Tensions and the Future of AI Policy

Despite the resolution regarding Mythos 5, the relationship between Anthropic and the Trump administration has been complex. Earlier in 2026, a significant dispute arose with the Department of Defense (DoD) over Anthropic's refusal to allow its Claude models to be used for mass domestic surveillance or fully autonomous weapons systems. This disagreement led to the DoD designating Anthropic as a "supply-chain risk to national security" and a directive for federal agencies to cease using Anthropic's technology, prompting various departments to phase out Claude from their workflows.

This broader conflict, which saw a federal judge block most of the DoD's "supply-chain risk" designation, highlights fundamental differences in philosophy regarding AI usage and ethical boundaries. Anthropic has consistently advocated for responsible AI development and deployment, with a "constitutional AI" approach that aims to embed safety and ethical considerations into its models.

The recent decision on Mythos 5, while a positive step for its availability, does not entirely erase these underlying tensions. It suggests a pragmatic approach by the government to allow access to critical AI tools for national interests while continuing to assert control over their deployment and potential misuse. The competitive landscape for federal AI contracts remains intense, with companies like OpenAI also releasing their latest models, such as GPT-5.6, to government-approved partners on the same day as the Mythos 5 announcement.

As AI capabilities continue to advance at an unprecedented pace, the interplay between technological innovation, corporate ethical stances, and government regulatory oversight will undoubtedly become more pronounced. The case of Anthropic's Mythos 5 serves as a clear illustration of this complex and evolving dynamic, setting precedents for how powerful AI will be managed and integrated into critical national infrastructure and enterprise operations moving forward.

Frequently Asked Questions

What is Anthropic Claude Mythos 5?

Anthropic Claude Mythos 5 is a powerful artificial intelligence model developed by Anthropic, specifically designed for advanced applications in cybersecurity, such as finding software vulnerabilities, and for scientific research, including generating complex hypotheses in fields like biology and healthcare.

Why was Anthropic Mythos 5 initially restricted by the Trump administration?

The Trump administration imposed export controls on Mythos 5 due to national security concerns. These included reports of unauthorized access to the model by entities potentially linked to China, as well as fears that the model could be "jailbroken" or misused for malicious purposes.

What does the lifting of the block mean for US companies and agencies?

The lifting of the block means that over 100 approved US companies and government agencies are now authorized to use Anthropic's Claude Mythos 5. This access also extends to their foreign national employees and Anthropic's own foreign national employees, without requiring a special export license.

How does this decision relate to Anthropic's broader relationship with the US government?

While the lifting of the Mythos 5 block is a specific de-escalation, Anthropic has had ongoing tensions with the Trump administration, particularly the Department of Defense. These disputes have centered on Anthropic's refusal to allow its AI models to be used for mass domestic surveillance or autonomous weapons systems, leading to previous bans and designations as a "supply-chain risk." The Mythos 5 decision indicates a pragmatic agreement for this specific model, but the broader policy dialogue continues.