NSA Reportedly Preparing Anthropic’s Mythos AI for Offensive Cyber Operations

Anthropic has reportedly assigned a small group of engineers to the National Security Agency (NSA) to help the agency explore using its advanced cybersecurity AI model, Mythos, according to a Financial Times report citing anonymous sources.

The report says the engineers are assisting with specific applications of the model, although it remains unclear whether Mythos is currently being used in active cyber operations. The NSA is responsible for collecting intelligence through methods such as wiretaps, undersea communications networks, and corporate partnerships, as well as conducting offensive cyber operations against foreign adversaries.

The report follows earlier coverage by Axios, which in April claimed that the NSA was using Mythos despite a federal restriction on Anthropic technology. The restriction reportedly stemmed from a Department of Defence decision to classify Anthropic as a supply-chain risk after the company refused to allow its AI systems to be used for mass domestic surveillance and autonomous weapons programs.

The latest development comes as governments increasingly seek access to Mythos, a model Anthropic has previously limited due to concerns that its cybersecurity capabilities could be exploited to identify software vulnerabilities and carry out cyberattacks.

An NSA spokesperson declined to confirm or deny the report, while Anthropic did not respond to requests for comment.