NTIA Suggests Government Refrain From Immediate Restrictions on AI

learning. Open AI models are open-source, allowing public access to data, while closed models are private. NTIA gathered public comment on the benefits and risks of open and closed models in response to President Joe Biden’s executive order on AI. Current evidence isn’t “sufficient to definitively determine either that restrictions on such open-weight models are warranted, or that restrictions will never be appropriate in the future,” NTIA said in its Report on Dual-Use Foundation Models with Widely Available Model Weights. The agency recommended the federal government “actively monitor a portfolio of risks that could arise from dual-use foundation models with widely available model weights and take steps to ensure that the government is prepared to act if heightened risks emerge.” However, NTIA laid out possible restrictions for the technology, including a ban on “wide distribution” of model weights, “controls on the exports of widely available model weights,” a licensing framework for access to models and limits on access to application programming interfaces and web interfaces. NTIA noted that restrictions on open public model weights “would impede transparency into advanced AI models.” Model weight restrictions could limit “collaborative efforts to understand and improve AI systems and slow progress in critical areas of research,” the agency said. Open Technology Institute Policy Director Prem Trivedi said in a statement Tuesday that NTIA is correct in recommending the rigorous collection and evaluation of empirical evidence, calling it the right starting point for policymaking on the issue.