Anthropic employees reportedly aid NSA in offensive AI deployment using Mythos Preview (heise.de)

• Anthropic engineers on-site at NSA to adapt Mythos Preview for offensive cyber use, per FT.
• Cooperation contradicts Pentagon ban on Anthropic technology after earlier conflict.
• Mythos Preview model designed for vulnerability discovery and attack path identification.

"Multiple Anthropic engineers are working on-site at the NSA to adapt the Mythos Preview AI model for offensive cyber operations, according to the Financial Times. The collaboration contradicts the public rift between Anthropic and the Pentagon, which had banned the company's technology from its systems. Anthropic justifies the cooperation by citing the need for strong offense to build strong defense. The exact scope of the engineers' work remains unclear."