Anthropic embeds engineers at the NSA for offensive cyber AI operations while seeking a pause on AI development ahead of a potential $1 trillion IPO.
Anthropic has stationed approximately six engineers at the National Security Agency to deploy its Mythos AI model for offensive cyber operations. The model, restricted to vetted partners like Microsoft and Apple, may target networks in China and Iran, though active involvement remains unconfirmed.
The company is simultaneously pursuing an IPO that could surpass a $1 trillion valuation. Anthropic has also called for a global pause on AI development, citing risks of recursive self-improvement. Earlier this year, the Pentagon labeled Anthropic a supply-chain risk after a $200 million contract collapsed over autonomy and surveillance concerns.
Anthropic’s dual role—supporting U.S. cyber operations while advocating for AI restraint—highlights tensions between commercial ambitions and ethical safeguards in the AI sector.