The U.S. Department of Commerce issued a report on Monday supporting open-weight generative artificial intelligence models, such as Meta's Llama 3.1. However, the report also recommended that the government develop "new capabilities" to monitor these models for potential risks.
The report, prepared by the National Telecommunications and Information Administration (NTIA) of the Department of Commerce, highlights that open-weight models expand the accessibility of generative AI to small businesses, researchers, non-profit organizations, and individual developers. For this reason, it suggests that the government should not impose restrictions on these open models without first assessing whether such limitations could negatively affect the market.
Benefits of Open AI
The document reflects similar views recently expressed by Federal Trade Commission (FTC) Chair Lina Khan, who believes that open models can enable more small businesses and entrepreneurs to bring their ideas to market, thus fostering healthy competition. Alan Davidson, Under Secretary of Commerce for Communications and Information and Administrator of NTIA, emphasized the importance of open AI systems in an official statement, noting that these models have the potential to influence competition and innovation, as well as the inherent risks of these advanced technologies.
Davidson commented that "the openness of the largest and most powerful AI systems will impact competition, innovation, and risks in these revolutionary tools." The NTIA report recognizes the importance of open AI systems and calls for more active monitoring of the risks associated with the broad availability of large-scale models. According to the report, the government has a fundamental role in supporting AI development while also creating capacities to identify and address emerging new risks.
Regulations and Future
This report comes amidst growing regulatory debates both nationally and internationally. For example, the state of California is considering SB 1047, which would require companies to improve their cybersecurity and develop control mechanisms for models using more than 1026 FLOP of computing power. Internationally, the European Union has established new requirements under its AI Act, focusing on copyright, transparency, and AI applications. Meta has expressed that EU policies could limit its ability to launch certain open models in the future. Meanwhile, several startups and major tech companies have voiced opposition to the proposed California legislation, arguing that these measures are too restrictive.
Although NTIA's stance is not entirely laissez-faire, the report urges the government to develop an ongoing program to evaluate the risks and benefits of open AI models. This would include researching the security of various AI models, supporting risk mitigation, and establishing specific risk indicator thresholds to guide potential policy changes.
Gina Raimondo, U.S. Secretary of Commerce, indicated that these measures align with President Joe Biden's executive order on AI. Raimondo emphasized that the Biden-Harris Administration is committed to maximizing the benefits of AI while minimizing its risks, ensuring that the report provides a roadmap for responsible AI innovation and U.S. leadership in this field.