GSA's draft AI procurement rule has improved but needs further reforms, contractors say

Contractors and experts told General Services Administration officials during a Tuesday listening session that the agency's revised draft rule for large language model procurements has improved but still needs further modifications. GSA published a draft "Basic Safeguarding of Artificial Intelligence Systems" rule in March before releasing a revised version titled the "Basic Safeguarding of Data within Large Language Model Artificial Intelligence Systems (LLMs)" last month. The proposed rule would require contractors to follow data protection and intellectual property safeguards when using large language models to process government data. The latest draft softened language around foreign AI components, assigned flowdown requirements based on four industry roles and removed a specific prohibition of diversity, equity and inclusion ideology in favor of generally prohibiting ideological dogmas. Amy Benson, government affairs vice president for Science Applications International Corporation, said stakeholders asked for clarity, practicality and alignment with commercial norms and the updated draft shows meaningful movement in those directions. However, Megan Peterson, a senior vice president at the Information Technology Industry Council, said the definition of data remains overly broad and must align with commercial practices. She added the definition of data outputs is unclear and could create confusion. Tim LeMaster, vice president of federal engineering at Lookout, said GSA should clarify what incidental large language model functionality means. Contractors have until Aug. 3 to comment on the draft rule.