Is the Pentagon Allowed to Surveil Americans With AI?
WASHINGTON — A public standoff between the Department of Defense and Anthropic has spotlighted a long-unresolved legal question: whether U.S. law permits the government to use advanced AI for mass surveillance of American citizens. The dispute escalated after Anthropic refused Pentagon demands to license its models for “all lawful purposes,” while rival OpenAI amended its own contract to explicitly bar intentional domestic surveillance of U.S. persons.
The controversy, detailed in a MIT Technology Review investigation published March 6, 2026, comes more than a decade after Edward Snowden’s revelations about NSA bulk data collection. It underscores ongoing tensions over how commercial AI systems may be repurposed for national-security missions that skirt or test the boundaries of domestic surveillance laws.
Anthropic has drawn a hard line against two categories of military use: mass surveillance of Americans and lethal autonomous weapons. According to multiple reports, the Pentagon responded by moving to blacklist the company from certain government contracts after CEO Dario Amodei resisted language that could enable “mass domestic surveillance.” In contrast, OpenAI initially agreed to let the Pentagon use its systems for any lawful purpose, prompting criticism that the broad wording left the door open to domestic spying.
Contract Negotiations and Policy Shifts
Under its original agreement, OpenAI permitted the Defense Department to deploy its AI for “all lawful purposes.” The company later amended the deal to include explicit language stating its systems “shall not be intentionally used for domestic surveillance of U.S. persons and nationals,” in line with relevant laws, according to statements reported by The New York Times and MIT Technology Review.
OpenAI also negotiated technical guardrails intended to uphold its safety principles, though critics argue the changes came only after public pressure and still leave ambiguity about potential indirect or non-intentional uses. Anthropic, meanwhile, has maintained its refusal to compromise on the two red lines, even at the risk of losing substantial federal business.
The Pentagon has pushed for broad “all lawful purposes” language across AI contracts, arguing it needs flexibility to counter emerging threats. However, the lack of clear statutory boundaries for AI-enabled surveillance of Americans has fueled the feud and public debate.
Competitive and Ethical Landscape
The diverging approaches of OpenAI and Anthropic highlight a split in the AI industry over military collaboration. OpenAI’s willingness to revise its contract has drawn accusations of caving to Pentagon pressure, as reported by The Verge. Anthropic’s firmer stance has positioned it as more restrictive, but also potentially isolated from lucrative defense contracts.
The episode revives long-standing questions about the intersection of commercial AI development and government surveillance powers. Legal experts note that while laws such as the Foreign Intelligence Surveillance Act and Executive Order 12333 impose restrictions, the precise application to modern AI tools capable of analyzing vast datasets remains unsettled.
Impact on Developers, Users and the Industry
For AI companies, the dispute signals increasing government pressure to loosen usage restrictions in exchange for access to federal markets. Developers must now weigh ethical commitments against commercial and national-security demands.
Users and civil-liberties advocates worry that permissive contracts could accelerate the deployment of powerful AI for domestic monitoring without adequate oversight. The industry as a whole faces heightened scrutiny over whether self-imposed safeguards can meaningfully constrain government use of dual-use technology.
What’s Next
The Pentagon has not publicly detailed its next steps regarding Anthropic’s blacklisting or whether additional AI providers will face similar demands. OpenAI’s amended contract language has not been fully released, leaving questions about the strength of its surveillance prohibitions.
Lawmakers and regulators are expected to face growing calls for clearer statutory guidance on AI and domestic surveillance. No timeline has been announced for potential legislation or updated Defense Department policies. Both OpenAI and Anthropic have declined to comment beyond previously reported statements.
The episode illustrates how rapidly evolving AI capabilities are colliding with decades-old legal frameworks originally designed for different technologies, ensuring the debate over Pentagon use of commercial AI for potential domestic surveillance will continue.
(Word count: 748)