The National Security Agency told lawmakers it is spending billions of dollars in taxpayer funds this year evaluating advanced AI models, according to sources familiar with classified estimates. The price tag is significantly greater than previously known and has led lawmakers to believe a comprehensive AI regulatory system could cost tens of billions per year.

President Donald Trump has resisted calls for greater federal oversight of AI, but after high-profile hacking incidents, the NSA's Artificial Intelligence Security Center began testing frontier models for national security vulnerabilities. The high cost is intensifying debate about the government's role in reviewing AI models from powerful companies.

Prior proposals suggested far lower costs. The Congressional Budget Office estimated the AI Security and Innovation Act would cost roughly $20 million per year. Some tech leaders like Elon Musk suggest frontier labs should review each other's models, while AI safety experts reject self-policing. Anthropic and OpenAI have indicated openness to greater federal oversight and potentially paying for it.

Currently, NSA's AI testing is funded by classified portions of the federal national security budget. The exact amount spent remains unclear. The Pentagon declined to comment on NSA's AI spending, though the Defense Department's annual budget is close to $1 trillion.