(WASHINGTON)–The Center for Democracy & Technology applauds the conclusions of a report released today by the Commerce Department’s National Telecommunications and Information Administration (NTIA), which considers the risks and benefits of, and potential policy approaches to, openness in AI.
NTIA’s inquiry was particularly focused on widely-available open “foundation” models that are published along with the model “weights” these systems rely on to produce outputs. This report and the consultation process leading to it were required by the Biden Administration’s Executive Order on AI.
In a victory for advocates of AI openness, the NTIA has recommended that the President should not pursue new legal restrictions on the publication of open foundation models without clearer evidence that such models pose significant risks greater than those already posed by other technologies such as the internet or closed AI models.
This conclusion is consistent with the perspective that CDT has advocated throughout the proceeding: in public events, in an academic paper co-authored by the leadership of CDT’s AI Governance Lab, in a coalition letter to the Commerce Secretary that CDT co-organized with Mozilla, and in CDT’s own comments to the NTIA, which are cited repeatedly throughout the report.
CDT’s Senior Advisor on AI Governance Kevin Bankston says:
“CDT appreciates that the NTIA correctly concluded that there is not yet enough evidence of novel risks from open foundation models to warrant new restrictions on their distribution. This is especially true when considering the much clearer evidence of benefits from AI openness: to transparency, security, innovation, competition, and the availability of AI to historically marginalized communities. These benefits accrue to everyone in the ecosystem, from those who develop and deploy AI systems to everyday users and consumers.
“This outcome wouldn’t be possible without a broad coalition—from civil liberties to civil rights organizations, from progressive groups to more market-oriented groups, along with advocates for openness in both government and industry, a broad range of academic experts from law, policy, and computer science, and the traditional open source software community—who raised their voices to defend the right to publish AI models with open model weights. In doing so, they have helped to advance the ability of everyone, not just a handful of large companies offering closed AI models, to benefit from this powerful new technology.”