GSA seeks public comment on draft regulation for safeguarding data in LLM acquisitions

The General Services Administration (GSA) published a draft clause for the General Services Acquisition Regulation (GSAR) focused on basic safeguarding of data within Large Language Model Artificial Intelligence Systems (LLMs). The draft targets federal acquisitions of LLM technologies, aiming to establish baseline security and privacy requirements during procurement processes.

The agency announced a 60-day public comment period to gather stakeholder feedback before advancing to potential deviation or formal rulemaking. The notice emphasizes the complexity of addressing data safeguarding in AI systems and seeks input from industry, academia, and other interested parties to refine the proposed requirements.

This action aligns with broader federal initiatives to develop safeguards for AI technologies procured by government agencies. The draft clause reflects an early-stage regulatory effort to integrate AI-specific considerations into federal acquisition practices, particularly concerning data protection and secure handling of sensitive information within LLM deployments.

The Federal Register notice does not specify the exact safeguards or technical controls included in the draft clause, nor does it provide a timeline for finalization beyond the open comment period. The GSA’s approach suggests a phased development process, with further revisions expected based on public input and evolving AI risk management frameworks.