Reports indicate the National Security Agency (NSA) is already using Anthropic's next-generation AI, 'Mythos Preview,' and that Claude is still widely used within the Department of Defense despite being designated a supply chain risk.
by
It has been reported that the U.S. National Security Agency (NSA) has already begun using Anthropic's next-generation high-performance AI model, ' Mythos Preview .' On the other hand, Anthropic has also been identified by the Department of Defense as a supply chain risk, and there have been moves within the administration to suspend or exclude its use. Nevertheless, due to the high necessity for cyber defense, the use of Anthropic's AI is once again being considered within the U.S. government.
NSA using Anthropic's Mythos despite Defense Department blacklist
https://www.axios.com/2026/04/19/nsa-anthropic-mythos-pentagon
Anthropic and Trump: Is a truce near? - POLITICO
https://www.politico.com/news/2026/04/17/anthropic-and-trump-is-a-truce-near-00879655
Anthropic's relationship with the Trump administration seems to be thawing | TechCrunch
https://techcrunch.com/2026/04/18/anthropics-relationship-with-the-trump-administration-seems-to-be-thawing/
Anthropic CEO Dario Amodi stated that the company would not allow its AI to be used for large-scale surveillance of American citizens or for the operation of fully autonomous weapons. However, Secretary of Defense Pete Heggesses's side demanded that the government have the final say and be able to use it for any legitimate purpose, leading to a breakdown in negotiations.
President Trump claims Anthropic's left-wing fanatics attempted to control the U.S. military and orders severance of ties; Secretary of Defense Heggesses to designate Anthropic as a 'supply chain risk' - GIGAZINE
Subsequently, President Donald Trump ordered federal agencies to cease using Anthropic's AI, and the Department of Defense designated Anthropic as a national security supply chain risk. This designation is a strong measure typically used for companies linked to foreign adversaries, and it could potentially force not only Anthropic's approximately $200 million (approximately 32 billion yen) contract with the Department of Defense, but also contractors working with the Department of Defense to cease using Anthropic products. Anthropic has indicated its intention to fight this measure in court, claiming it is unfair.
AI company Anthropic officially designated as a 'supply chain risk to U.S. national security,' Anthropic declares it will fight the case in court - GIGAZINE
However, what changed the dynamic between Anthropic and the US government was Anthropic's next-generation AI model, 'Mythos.' Anthropic claims that this model has hacking capabilities far superior to conventional AI, capable of autonomously finding and exploiting complex software vulnerabilities and zero-day vulnerabilities, as well as autonomously executing end-to-end cyberattacks by connecting multiple attack methods while moving within a company's IT systems.
Anthropic develops 'Claude Mythos Preview,' an AI with extremely high cyberattack capabilities, and has also launched 'Project Glasswing,' which will provide a preview version to Microsoft, Apple, and others - GIGAZINE
Given these risks, Anthropic explains that it is initially providing access to a limited number of companies, including JP Morgan, Amazon, and Apple, with the aim of helping them find critical code flaws before state-sponsored hackers and cybercriminals get ahead of them. According to the overseas media outlet Axios, the NSA, which is under the supervision of the Department of Defense, is among the organizations that have been given early access to Mythos Preview, suggesting that the government's cybersecurity needs are outweighing the conflict between the Department of Defense and Anthropic. It is unclear how Mythos Preview is being used within the NSA, but other government organizations are reportedly using AI primarily to scan their own environments and detect exploitable security vulnerabilities.
At the same time, federal government organizations that have high hopes for Anthropic's AI capabilities are reportedly taking steps to circumvent the Trump administration's tough stance against Anthropic in practical terms. According to the overseas media outlet Politico, the Commerce Department's Center for AI Standards and Innovation and the Treasury Department, which is seeking to investigate the impact on banks' cyber defenses, have requested access to Mythos, and government agencies are urgently exploring ways to use the model to assess its potential dangers and defensive usefulness.
by Anthropic
Furthermore, according to a source within the administration, 'almost every agency' except the Department of Defense is eager to use Anthropic's technology, highlighting a difference in enthusiasm between the U.S. government as a whole and the Department of Defense.
In fact, the Office of Management and Budget (OMB) has explained in emails sent to several federal agencies that it is considering whether it can provide a 'revised version' of Mythos to government agencies. Gregory Barbacia, the OMB's chief information officer, said they are 'working with model providers, industry, and intelligence agencies to see if it is feasible to provide it with sufficient guardrails and safety measures in place,' and indicated that more information will be released in the coming weeks.
Sean Cairncross, director of the Office of the National Cyber Director, held a conference call with private companies and, with the backing of White House Chief of Staff Susie Wiles and Vice President J.D. Vance, is reportedly leading the administration's response to this issue. During the meeting, Mythos was praised as 'a symbol of American AI innovation,' but warnings were also made about the potential dangers to corporate networks.
Amidst these developments, signs of a thaw in relations between Anthropic and the U.S. government began to appear. In April 2026, CEO Amodi met with Chief of Staff Wiles, Treasury Secretary Scott Bescent, and Director Cairncross, and both the U.S. government and Anthropic described the meeting as a 'productive starting point.'
The U.S. government stated that they discussed opportunities for cooperation and ways to address challenges associated with technological expansion, while Anthropic said they also discussed ways to collaborate with the government on common issues such as cybersecurity, America's AI competitiveness, and AI safety.
by World Economic Forum
Jack Clark, co-founder of Anthropic, described the dispute over supply chain risk designation as a 'contractual issue in the narrow sense' and stated that the company will continue to explain its latest model to governments.
This incident demonstrates how the government's policy of shutting out companies on the grounds of AI's national security risks has been challenged by the powerful cyber capabilities of AI itself. Anthropic is still in legal conflict with the Department of Defense, but the government views Mythos as a dangerous entity while simultaneously recognizing it as a technology that cannot be ignored for national defense and the protection of financial infrastructure.
Related Posts:
in AI, Posted by log1i_yk