- The report argues that GenAI development and open-source diffusion are advancing rapidly, while U.S. governance mechanisms move more slowly and must adapt without waiting for perfect information.
- In the next 12 to 18 months, the report identifies synthetic media and AI-enabled disinformation as pressing threats to trust in democratic institutions during the 2024 election cycle.
- The report recommends that NIST convene industry on a voluntary synthetic media code of conduct for elections and that Congress assign a lead agency to alert the public about synthetic media in federal elections.
- The report calls for sector regulators to apply existing authorities to GenAI, gain stronger in-house AI expertise, and focus oversight on highly consequential uses with significant benefits or harms.
- The report warns that foreign digital platforms from countries of concern may amplify malign content and sensitive-data risks as they converge with GenAI, and recommends both targeted restrictions and a broader risk-based framework.
This Special Edition Report frames governance of generative AI as an urgent national security and societal challenge for the United States. It opens with a primer describing GenAI as a class of systems that generate text, images, audio, code, and other outputs from natural language or multimedia inputs, while noting limitations such as incorrect outputs and dependence on flawed training data. The governance memo argues that technological development and open-source diffusion are moving faster than governance mechanisms, creating risks for democratic institutions, elections, privacy, fairness, and international stability. Its principal policy focus is near-term action: protect U.S. digital information and election systems from synthetic media and disinformation, help regulators identify highly consequential AI uses by sector, address foreign digital platforms from countries of concern, consider a centralized AI authority over time, and create a G20-based Forum on AI Risk and Resilience. The appendix develops the FAIRR proposal as a soft-law, multilateral and multi-stakeholder mechanism involving states, regulators, private firms, academia, and civil society. The report’s policy relevance lies in its call for the U.S. government to act while learning, using existing authorities and expertise-building while developing new domestic and international governance tools.
GenAI opens a range of new capabilities and applications that will shape the geopolitical balance, changing the calculus for military, diplomatic, and economic power.
SCSP
- Protect digital information and elections systems with a synthetic media code of conduct. NIST should convene industry to agree to a voluntary standard of conduct for synthetic media around elections.
- Scale public digital literacy education and disinformation awareness. Congress should authorize a federal entity to alert the public of synthetic media use in federal elections.
- Require content distribution platforms to technically support a content and provenance standard. A trusted federal entity should monitor platforms and enforce transparency levers to identify whether content is GenAI generated or modified.
- Focus regulatory efforts on AI use cases that have highly consequential impacts on society. Regulators should focus on encouraging AI that has significant benefits and mitigating the worst of the harms.
Published by the Special Competitive Studies Project. Co-publishers and external authors, where they exist, are credited in the byline above.