OpenAI has reopened talks with the Department of Defense to amend a hastily struck contract and add explicit limits on domestic surveillance, CEO Sam Altman told employees. In an internal memo Altman said the original agreement “looked opportunistic and sloppy” and that new language will bar intentional use of OpenAI systems for domestic surveillance of U.S. persons and nationals. OpenAI’s national-security lead, Katrina Mulligan, also said the renegotiation would block certain Defense Intelligence Components from using OpenAI services without separate contract changes. The company framed the changes as a technical and contractual safety commitment, pointing to the Fourth Amendment, the National Security Act and FISA as constraints on permissible uses. Legal and policy analysts welcomed tighter language but raised questions about enforceability and commercial data-use boundaries. The development follows public backlash over AI firm contracts with the Pentagon and growing scrutiny over whether commercial models should be used in military or surveillance contexts. Why it matters to campuses: research partnerships, sponsored computing time, and classified-government collaborations could face new contractual guardrails as companies and agencies redraw terms. University research offices and tech-transfer teams should watch for changes that could affect procurement, export-control assessments, and student or faculty access to commercial models.