The U.S. National Security Agency Uses the "Anthropic" Tool Despite Being Classified as a Security Threat

The Use of the U.S. National Security Agency's “Mythos Preview” Tool from Anthropic

The U.S. National Security Agency (NSA) is using the artificial intelligence tool “Mythos Preview” from the company “Anthropic,” despite the U.S. Department of Defense (Pentagon) officially classifying the company as a risk to supply chains, according to Axios on Sunday.

The report noted that this tool is being used more widely within the U.S. Department of Defense, reflecting an expansion in the adoption of artificial intelligence technologies within security institutions.

Lack of Official Comment from Involved Parties

So far, no official statements have been issued by “Anthropic,” the U.S. National Security Agency, or the Department of Defense regarding this information, while the NSA is part of the Pentagon.

This silence comes amid increasing debate over the use of artificial intelligence technologies within U.S. military and security institutions.

Discussions Between the Trump Administration and “Anthropic” After Previous Dispute

According to a report from Reuters, the administration of President Donald Trump discussed with the CEO of “Anthropic” last week the possibility of cooperation between the two parties, marking the first time since the dispute earlier this year between the company and the Department of Defense regarding how to use its artificial intelligence models.

This rapprochement indicates a reopening of dialogue channels despite previous tensions between the two parties.

Security Concerns Over the “Mythos” Model's Capabilities

These developments come amid growing concerns that the “Mythos” model, the latest from “Anthropic,” could enhance cyber attack capabilities, especially as its programming and semi-autonomous functioning abilities evolve.

The company previously stated that the model is “the most advanced yet in programming and AI assistant tasks,” referring to its ability to perform complex tasks in a semi-autonomous manner.

Experts Warn of Exploiting Security Vulnerabilities

Cybersecurity experts have clarified that the advanced capabilities of this model in programming may grant it unprecedented potential to discover security vulnerabilities in digital systems, as well as develop potential methods to exploit them.

This raises concerns about the use of artificial intelligence technologies for offensive or illicit purposes if not strictly regulated.