Personnel from the National Security Agency (NSA) are reportedly accessing Mythos Preview, an AI system noted for its advanced capabilities in computer security tasks. According to two sources who spoke to Axios, the model is proficient in identifying vulnerabilities in legacy code and can autonomously devise methods to exploit these weaknesses, often surpassing human hackers in various cybersecurity challenges.
This development comes in direct conflict with a decision made by the Department of Defense (DoD) in March, which labeled Anthropic, the developer behind Mythos, as a supply chain risk and attempted to exclude it from federal contracts. The Pentagon’s actions followed Anthropic’s refusal to relax its safeguards concerning autonomous weapons and surveillance technologies.
Meanwhile, White House officials are reportedly seeking a new approach to leverage this technology. Last week, Dario Amodei, CEO of Anthropic, met with key government figures, including Treasury Secretary Scott Bessent and White House Chief of Staff Susie Wiles, to discuss the use of Mythos and broader safety practices. The White House characterized the meeting as “productive and constructive.”
The urgency surrounding Mythos underscores its critical role in national defense, even amid rising political tensions. Currently, only about 40 organizations globally have obtained access to Mythos, primarily due to concerns regarding its offensive capabilities. Those entities with clearance are using it to scan their own systems for exploitable vulnerabilities.
According to court records, Anthropic has been involved in high-level military operations since 2024. The company is currently suing federal agencies over the supply chain risk designation, arguing that the label is retaliatory, stemming from its refusal to grant unrestricted Pentagon access due to concerns about potential mass domestic surveillance.
While a federal appeals court recently denied Anthropic’s request to temporarily block the designation, many agencies continue to utilize its tools under existing agreements. This situation highlights how military AI policies are increasingly shaped by procurement contracts rather than formal laws or regulations, resulting in conflicting standards among different branches of the military.
In a previously issued directive, former President Donald Trump instructed all federal agencies to cease using Anthropic’s technology. However, a federal judge temporarily blocked the enforcement of that directive in March, labeling the company as run by “left wing nut jobs” attempting to “strong arm” defense officials. When asked about Amodei’s recent visit to the White House, Trump stated he had “no idea” about the meeting.
The ongoing dynamics concerning Mythos and Anthropic signal an evolving landscape in the intersection of AI technology and national security. As government entities navigate the complexities of integrating advanced AI systems while addressing ethical considerations, the broader implications for cybersecurity and defense strategies continue to unfold.
See also
Anthropic’s Claims of AI-Driven Cyberattacks Raise Industry Skepticism
Anthropic Reports AI-Driven Cyberattack Linked to Chinese Espionage
Quantum Computing Threatens Current Cryptography, Experts Seek Solutions
Anthropic’s Claude AI exploited in significant cyber-espionage operation
AI Poisoning Attacks Surge 40%: Businesses Face Growing Cybersecurity Risks
