The National Security Agency (NSA) is reportedly deploying Mythos Preview, a powerful artificial intelligence model that its developer, Anthropic, deemed too dangerous for public release due to its “offensive cyberattacks” capabilities. This deployment occurs weeks after the Department of Defense, the NSA’s parent agency, labeled Anthropic a “supply-chain risk” for refusing to grant Pentagon officials “unrestricted access” to its AI models, specifically for “mass domestic surveillance and autonomous weapons development.” This situation reveals a critical internal conflict within the national security apparatus regarding the control and deployment of advanced surveillance technology, bypassing stated concerns about national security and public safety.
Anthropic announced Mythos “earlier this month” as a frontier model specifically designed for cybersecurity tasks. Despite its stated purpose, the company explicitly stated that the model was “too capable of offensive cyberattacks to be released publicly,” leading to severe restrictions on its distribution.
As a direct consequence of these capabilities, Anthropic limited access to Mythos to approximately 40 organizations globally. Of these, the company has publicly named only a dozen, leaving the vast majority of recipients undisclosed to the public.
The NSA appears to be among these undisclosed recipients, gaining access to a technology deemed too dangerous for general release. This access raises fundamental questions about the transparency and oversight of powerful AI tools within government agencies and the potential for elite capture of critical national security functions.
Further underscoring the transnational nature of this elite technological control, the U.K.’s AI Security Institute has also confirmed its access to Mythos, indicating a coordinated, international approach to managing or deploying such sensitive AI outside of direct national democratic accountability.
Elite Control and National Security Bypass
The NSA is said to be using Mythos primarily for scanning environments for exploitable vulnerabilities, a task that could have broad implications for both national defense and domestic security. The deployment of a model with known offensive capabilities by a domestic intelligence agency highlights the expanding reach of the surveillance state over the native population.
This expanding use of Anthropic’s tools by the U.S. military occurs even as the military simultaneously argues in court that these very tools can threaten national security. This contradiction exposes a deep rift or a deliberate bypass of established security protocols within the defense establishment, prioritizing access to advanced tech over stated national security risks.
The Pentagon’s initial dispute with Anthropic originated from the company’s refusal to make its Claude model available for “mass domestic surveillance and autonomous weapons development.” This refusal by a private corporation to fully cede control of its technology to the state underscores the growing power of technocratic elites over national security prerogatives, effectively dictating terms to sovereign governments.
The Price of Unchecked Power
The implications of a private entity dictating the terms of access for tools capable of “mass domestic surveillance” are profound for the native population. It suggests a future where the privacy and autonomy of citizens could be compromised by technologies developed and controlled outside direct democratic accountability, contributing to a managed decline of individual liberties.
The NSA’s access to Mythos also coincides with what appears to be a thawing relationship between Anthropic and the current administration. This development suggests political maneuvering at the highest levels to secure access to these critical technologies, potentially overriding previous security concerns and further entrenching elite collaboration.
“Last Friday,” Anthropic chief executive Dario Amodei met with White House chief of staff Susie Wiles and Secretary of the Treasury Scott Bessent. Such high-level meetings between private tech executives and top government officials signal the deep integration of corporate interests into national policy-making, often shielded from public scrutiny.
The White House reportedly called the meeting productive, indicating a potential resolution to the previous disputes and further solidifying the influence of private tech firms over state capabilities, thereby accelerating the transfer of sovereignty from the people to unelected technological and political elites.
When contacted for comment by TechCrunch, Anthropic declined to provide any statement, maintaining a veil of secrecy around its operations and its engagements with national security agencies. This lack of transparency further obscures the true extent of elite collaboration and its impact on national sovereignty and the future of the native working class.