As AI developments bring new policy considerations to the forefront, TFS’s workstream on US AI Governance aims to inform and contribute to federal efforts toward AI governance. Ultimately, our work helps shape the rules—enforceable mechanisms and voluntary frameworks—needed to ensure that advanced AI systems are developed in a safe, ethical, and responsible manner.

The US, home to many leading industry and academic AI developers, has a unique opportunity to steer the development and deployment of AI systems. To date, US federal government entities have drafted horizontal voluntary principles and governance frameworks for how AI should be developed. The combination of such approaches is meant to guide developers toward exercising caution while establishing concrete safeguards for a limited range of industrial applications.  Although useful, such a light-touch approach is unlikely to suffice in the long term, as advanced AI systems present novel threats to economic security, democratic stability, public safety, and the administration of law. Regulations will be needed to mitigate risks and to hold accountable those that transgress legal or regulatory boundaries.

The good news is that there are emerging opportunities to put the right rules in place. From the Biden Administration’s meetings with industry leaders to Congressional testimonies, the FTC’s ongoing guidance, a suite of proposed bills in Congress, and a number of agencies requesting comments, there is growing momentum for regulatory action. Yet, having yet to see a comprehensive piece of legislation in the US, we believe this to be a relatively neglected policy area. We must capitalize on this window of opportunity to enact meaningful and effective governance mechanisms that ensure the responsible development and deployment of AI systems.

Such mechanisms ought to build upon foundational frameworks, such as the OECD Principles on AI, to which the US has already agreed, and the Blueprint for an AI Bill of Rights. Furthermore, these mechanisms should be co-shaped by a  broad range of civil society actors because regulation is more likely to serve the public interest when it is informed by diverse perspectives. Moreover, regulatory frameworks ought to harmonize with those of US allies: the US should seek to complement (and not compete with) regulatory efforts of the EU and the UK. Finally, we believe global coordination will be key. AI is a global phenomenon, and addressing its challenges and opportunities will require a coordinated international approach to ensure safe and ethical outcomes.

Our work thus far has included:

• A response to the National Telecommunications and Information Administration’s (NTIA) Request for Comment, in which we emphasized the need for scrutiny in the design and development of general-purpose AI systems (GPAIS).
• A response to the U.S. Office of Science and Technology Policy (OSTP) Request for Information on National Priorities for AI, which put forward national priorities focused on security standards, measurement and evaluation frameworks, and an industry-wide code of conduct for GPAIS development.
• A submission to the NIST Generative AI Public Working Group Request for Resources, in which we provided a list of clauses to govern the development of GPAIS.