NSA Accesses Anthropic Model After Pentagon Blacklist
The NSA has accessed Anthropic's Mythos Preview AI model despite the Pentagon designating Anthropic as a supply chain risk due to disagreements over safeguarding controls related to autonomous weapons and surveillance. This indicates internal federal division on the use of Anthropic technology, with the NSA utilizing the model primarily for security testing despite official restrictions. Anthropic had limited Mythos access to about 40 organizations because of its offensive cyber capabilities.
TLDR The NSA reportedly used Anthropic’s Mythos Preview despite a Pentagon supply chain risk designation. Axios cited sources who confirmed the NSA accessed the model after the restriction. The Pentagon labeled Anthropic a supply chain risk in March over safeguard disputes. The dispute involved Anthropic’s refusal to loosen controls tied to autonomous weapons and surveillance. Anthropic limited Mythos Preview access to about 40 organizations due to offensive cyber capabilities. The National Security Agency has used Anthropic’s Mythos Preview despite a Pentagon supply chain risk designation. Axios cited multiple sources who confirmed the agency’s access to the model. The report shows internal differences across federal departments over Anthropic technology use. Anthropic Model Access Expands Inside NSA Axios reported that two sources confirmed NSA use of Mythos Preview. One source said the department expanded usage beyond limited internal testing. However, the Pentagon designated Anthropic as a supply chain risk in March. The designation restricted the company’s technology in military contracts after a policy dispute. The dispute centered on Anthropic’s refusal to loosen safeguards. The company declined to adjust controls linked to autonomous weapons and domestic surveillance systems. As a result, the Pentagon limited procurement pathways for Anthropic products. Still, Axios reported that the NSA accessed Mythos through existing channels. The report did not explain the technical route used for access. However, it confirmed that usage occurred after the risk designation. Axios stated that it remains unclear how the NSA deployed the model. Yet, other authorized organizations have used Mythos to scan systems for vulnerabilities. Anthropic limited Mythos Preview access to about 40 organizations. The company restricted access because of the model’s offensive cyber capabilities. Sources told Axios that the NSA relied on the model’s security testing features. They did no...
Comments
Log in to comment