Key Points
- Federal officials have rejected Anthropic’s proposal to distribute its Mythos AI system to an additional 70 organizations, which would total approximately 120 entities.
- The company withheld Mythos from public distribution due to its sophisticated capability to identify and leverage security weaknesses in software systems.
- A select consortium including Apple, Microsoft, and Nvidia currently maintains access through an initiative named “Glasswing.”
- Federal authorities express concern that Anthropic’s computational infrastructure is insufficient to accommodate expanded distribution without compromising government usage.
- Unauthorized individuals gained entry to Mythos on the identical day the company revealed its controlled distribution strategy.
In early April 2026, Anthropic introduced Mythos, characterizing it as an artificial intelligence system with the capability to identify and take advantage of weaknesses in essential software infrastructure. According to the company’s statements, the model could uncover security gaps that remained hidden for decades, evading detection by both human security professionals and automated scanning systems.
Given these powerful abilities, Anthropic chose to withhold Mythos from widespread public availability. The company instead initiated a restricted program dubbed “Glasswing,” providing a specialized version of the system to a carefully selected group of corporations, such as Apple, Microsoft, and Nvidia, with the goal of strengthening their cybersecurity frameworks.
Anthropic subsequently put forward a proposal to extend access to approximately 70 more entities, raising the overall count of organizations utilizing Mythos to roughly 120. This expansion initiative has encountered resistance from federal authorities.
Administration Voices Infrastructure and Security Objections
A White House representative disclosed to Bloomberg News that officials within the Trump administration have communicated their disapproval of the expansion strategy to Anthropic. The representative requested anonymity when providing this information.
A primary issue involves computational resources. Federal officials are apprehensive that distributing Mythos to additional corporations might diminish the government’s capacity to utilize the technology efficiently, as reported by the Wall Street Journal.
A White House spokesperson informed the Journal that the administration is attempting to strike a balance between fostering innovation and maintaining security protocols while guaranteeing responsible deployment of artificial intelligence systems.
This opposition intensifies an already tense dynamic between the Trump administration and Anthropic. In February, President Trump directed federal agencies to discontinue use of Anthropic’s products after Defense Secretary Pete Hegseth designated the organization as a national security supply chain vulnerability.
Anthropic is presently contesting these directives through legal proceedings.
Recent Discussions and Breach Incident
The relationship between Anthropic and federal authorities demonstrated modest improvement in early April. CEO Dario Amodei participated in discussions with government representatives at the White House, with Anthropic characterizing these exchanges as “productive.”
However, complications arose when Bloomberg News disclosed that a limited number of unauthorized individuals within a private digital community obtained access to Mythos on the very day Anthropic initially publicized its restricted distribution approach. Anthropic acknowledged it is examining the breach.
Several observers have suggested Anthropic may be exaggerating Mythos’s actual capabilities. The organization competes in an intensely competitive sector, with competitor OpenAI pursuing comparable technology development.
Anthropic refused to provide commentary regarding the White House’s resistance. The White House did not supply a response to inquiries for comment.
