According to @JTillipman, AI vendors can and regularly do restrict U.S. government use of their models through specific acquisition pathways, license terms, and data rights clauses, as reported on her explainer at jessicatillipman.com. According to Jessica Tillipman (GW Law), limits on government use hinge on the contract vehicle (e.g., commercial item acquisitions), the type of license (commercial licenses with usage caps or safety restrictions), and negotiated provisions like data rights, IP, and acceptable use, which can constrain Department of Defense deployments and mission profiles. As reported by Jessica Tillipman, agencies that accept standard commercial terms may be bound by vendor-imposed restrictions on model customization, fine-tuning, red-teaming access, and downstream use, affecting procurement timelines and compliance. According to @JTillipman, understanding FAR and DFARS data rights, click-through licenses, and other pathways creates business opportunities for AI companies to protect safety policies while selling to defense and civilian agencies, and for buyers to negotiate tailored rights for mission-critical applications.

According to Fox News AI, the Pentagon has issued an ultimatum to an artificial intelligence firm to lift contractual limits on military use by Friday or lose a $200 million deal, citing national security needs and operational flexibility (as reported by Fox News). The development signals rising demand for dual‑use AI tools in defense procurement and could reshape compliance terms for foundation models and model-as-a-service offerings across DoD programs, according to Fox News. For AI vendors, the near-term business opportunity lies in clarifying acceptable use policies, export controls, and deployment guardrails to meet defense accreditation while preserving safety commitments, as reported by Fox News.