The General Services Administration (GSA) has opened a public consultation on a draft Acquisition Regulation clause designed to safeguard government data processed through large language models (LLMs) and other artificial intelligence tools. This initiative reflects heightened federal attention on secure, ethical AI deployment in government contracting.
Issued in the Federal Register, the draft clause addresses key concerns including data protection, intellectual property rights, and ethical AI use. It is shaped by directives from a recent executive order on AI and guidance from the Office of Management and Budget. The provision targets the security, integrity, and handling of government data within AI systems, ensuring these standards apply broadly across GSA-managed government-wide contracts such as the Federal Supply Schedule and OASIS+.
GSA is soliciting comments by early August and will facilitate a public listening session mid-July to gather stakeholder perspectives. Input is sought on five core areas: the effectiveness of proposed clause changes; rules for government data ownership and contractor accountability; clarity regarding the roles of contractors, AI developers, system operators, integrators, and service providers; the implementation of downstream flowdown requirements; and risks linked to foreign ownership or control over LLMs that might compromise government data or decisions.
The agency notes that feedback will guide potential adjustments including deviations and formal rulemaking. This effort aligns with GSA’s wider acquisition modernization drive, which includes ongoing reforms like the Revolutionary FAR Overhaul, aimed at updating procurement systems to better support emerging technologies such as AI.