Is the U.S. Reconsidering Anthropic Ban After 'Mythos Shock'? Possible Reinstatement Under Review

by Oh Suyon

Published 17 Apr.2026 06:20(KST)

Updated 17 Apr.2026 08:02(KST)

The U.S. federal government, which had previously banned Anthropic’s model, now appears to be reconsidering its decision in light of what is being called the “Mythos shock.”

According to Bloomberg News on the 16th (local time), Gregory Bavasia, Chief Information Officer (CIO) of the Office of Management and Budget (OMB) at the White House, stated in an email sent to various departments on the 14th that the OMB is preparing protective measures so that government agencies can use Anthropic’s “Claude Mythos” model.

Anthropic Logo. Photo by Reuters and Yonhap News

In an email titled "Access to the Mythos Model," CIO Bavasia said, "We are working closely with the model provider, industry partners, and intelligence authorities to establish appropriate security protocols and safeguards before making a modified version of the model available to government agencies."

This email was sent to the Department of Defense (War Department), Department of the Treasury, Department of Commerce, Department of Homeland Security, Department of Justice, and the State Department, and included a note that more information will be provided in the coming weeks.

However, the email did not explicitly state that government agencies would be granted access to Mythos, nor did it provide a specific timeline for when access might be granted or how it would be utilized.

Bloomberg explained that this indicates the U.S. government remains interested in Anthropic’s tools, even amid ongoing public disputes between Anthropic and senior officials from the Trump administration. Citing recent sources, the outlet reported that the Treasury Department is seeking to gain access to Mythos. Previously, the Treasury Department suspended its use of Anthropic in accordance with instructions from President Donald Trump.

On April 7, Anthropic announced that its newly developed Mythos model possesses expert-level capabilities in detecting powerful security vulnerabilities. The company said it would first provide the model to major technology and financial firms before releasing it to the general public. Subsequently, the U.S. government, along with other governments and financial authorities around the world, held emergency meetings to assess the impact of artificial intelligence (AI) models like Mythos.

On this day, Anthropic also released "Claude Opus 4.7," a model with more limited cybersecurity capabilities compared to Mythos. This model, the most advanced version available to the public, has significantly enhanced abilities in areas such as coding and financial analysis compared to previous versions.