Is the Pentagon allowed to surveil Americans with AI?

A high-stakes feud between AI company Anthropic and the U.S. Department of Defense has ignited a critical debate: does existing law permit the government to use AI for mass domestic surveillance? The flashpoint was the Pentagon's desire to use Anthropic's Claude AI to analyze bulk commercial data on Americans, which Anthropic refused, citing ethical redlines against mass surveillance and autonomous weapons. In response, the Pentagon designated Anthropic a supply chain risk. Meanwhile, rival OpenAI initially signed a deal allowing Pentagon use for "all lawful purposes," sparking user protests and app deletions that forced OpenAI to publicly rework its contract to explicitly ban domestic surveillance and intelligence agency use.

Legal experts explain the core issue: a vast gap exists between what ordinary people consider surveillance and what current U.S. law regulates. The government can legally purchase commercial data—like mobile location and web browsing records—from the open market, a practice increasingly used by agencies from the FBI to the NSA. This data, combined with public information like social media posts, creates massive datasets that AI can supercharge for analysis. Laws like the Fourth Amendment and the Foreign Intelligence Surveillance Act (FISA) were written for a pre-internet era and don't adequately cover modern data collection or AI's analytical power. Anthropic CEO Dario Amodei argues the law hasn't caught up to AI capabilities, while OpenAI's Sam Altman suggested their contract simply needed to reference existing prohibitions, highlighting the profound legal uncertainty at the intersection of national security, privacy, and artificial intelligence.