Anthropic Regains Partial Access to Mythos 5 AI Model for Select Clients

• June 28, 2026

Anthropic, a prominent artificial intelligence research company, has achieved partial restoration of access to its Mythos 5 AI model for certain clients, following a recent government-mandated restriction. This development comes after the United States government ordered the suspension of foreign access to advanced AI models, including Anthropic’s Mythos 5 and Fable 5, critically impacting the company’s cybersecurity evaluation services.

Restricted Access to Advanced AI Models

Several weeks ago, the U.S. government implemented a directive to restrict foreign nationals’ access to cutting-edge AI models that have significant cybersecurity applications. The order affected multiple providers, including Anthropic, which had to comply by terminating access to these models across a wide client base. Due to technical limitations, Anthropic was initially unable to selectively restrict access and thus disconnected nearly all users from these AI-powered cybersecurity tools.

Mythos 5, one of Anthropic’s flagship AI models, plays a pivotal role in assessing and detecting cyber threats. Its suspension disrupted various operations that rely on the model’s advanced capabilities, prompting the company to seek ways to regain client access while adhering to the imposed restrictions.

Following negotiations and compliance efforts, Anthropic has now secured the authorization to restore Mythos 5 access, albeit on a partial basis. The reinstatement applies only to a select group of clients, reflecting a calibrated approach to balancing regulatory compliance with operational needs in the cybersecurity ecosystem.

This partial restoration marks an important step in navigating the complex regulatory landscape surrounding powerful AI technologies, particularly those with national security implications. It also underscores the challenges faced by AI companies in maintaining service continuity when regulatory actions impose broad and immediate access constraints.

At this time, details regarding the specific criteria for client selection and the scope of regained access have not been disclosed. The company’s ability to selectively enable access suggests advancements in their client management capabilities or the implementation of mechanisms to comply with granular regulatory demands.

The situation highlights the ongoing tension between technological innovation in AI and emerging efforts to control its distribution internationally. As governments seek to balance the benefits of AI against potential risks, companies like Anthropic are navigating unprecedented operational and legal frameworks.

Industry observers are closely monitoring how Anthropic and other AI developers adjust their strategies in response to government regulations that affect the availability of advanced models. The developments surrounding Mythos 5 access are likely to influence broader discussions on AI governance, cybersecurity, and international cooperation in technology controls.

Anthropic’s partial restoration of Mythos 5 access demonstrates a cautious path forward, reflecting both regulatory compliance and a commitment to serving key clients reliant on their technology. The company is expected to continue adapting as regulatory landscapes evolve and as demand for sophisticated AI cybersecurity applications grows.

Anthropic has secured limited reinstatement of Mythos 5 AI model access for some clients following a US government restriction.