AI Summit_Sept. 13 2024

Congressional Research Service

The Promoting Responsible Evaluation and Procurement to Advance Readiness for Enterprise-wide Deployment for Artificial Intelligence Act , or the PREPARED for AI Act (S. 4495) was introduced on June 11, 2024. The bill would require the Federal Acquisition Regulatory Council to update the Federal Acquisition Regulation to ensure that federal agency acquisition planning includes risk evaluation and impact assessments, with considerations of such factors as data ownership, privacy, and security; data and model assessment processes; scope of use; ongoing monitoring and audits; environmental impact; and adverse event reporting. The bill would prohibit federal agency procurement of AI for a high risk use case prior to establishing specified terms into relevant contracts, agreements, and employee guidelines. In addition to procurement provisions, the bill would require the Office of Management and Budget (OMB) to establish a Chief AI Officers Council to coordinate interagency governance of AI, and to develop procedures to ensure that any adverse incidents involving AI used by federal agencies is promptly reported to the agencies. )XUWKHU WKH ELOO ZRXOG UHTXLUH WKH KHDG RI HDFK IHGHUDO DJHQF\ WR XQGHUWDNH DFWLYLWLHV WR ³HQVXUH WKH responsible adoption of $, ´ to establish an AI governance board comprised of agency officials, and to develop a publicly-available risk classification system for agency use cases of AI (including, at a minimum, unacceptable, high, medium, and low risk classifications). Additionally, the bill would require each agency to designate a Chief AI Officer (CAIO) to develop and implement a process for identifying and evaluating risks posed by the deployment of AI in agency use cases. The bill would direct each agency to require deployers and developers of AI systems obtained by the agency to submit relevant documentation about the AI model, with additional documentation for medium and high risk use cases. The bill would also prohibit certain agency uses of AI, including for facial emotion recognition; using biometric data to deduce or infer certain characteristics such as race, political opinion, or sexual orientation; or for evaluating, rating, or scoring trustworthiness or social standing of an individual. The Future of Artificial Intelligence Innovation Act of 2024 (S. 4178) was introduced on April 18, 2024. In addition to multiple provisions focused on supporting AI R&D, the bill would require the Under Secretary of Commerce for Standards and Technology (who also serves as the Director of NIST) to establish an AI Safety Institute, the goals of which would include assisting the private sector and other federal agencies in developing voluntary best practices for assessing AI systems, and providing technical assistance for AI adoption and use across the federal government. The bill would also require NIST to establish an AI Safety Institute Consortium ² a consortium of public and private stakeholders with AI expertise ² to support the Institute in carrying out is functions. Additionally, the bill would require Department of Commerce (DOC), Department of State, and the Office of Science and Technology Policy 2673 WR VHHN WR IRUP DQ DOOLDQFH RU FRDOLWLRQ ZLWK ³OLNH - PLQGHG JRYHUQPHQWV RI IRUHLJQ FRXQWULHV´ on innovation, development, and harmonization of AI standards. The bill would direct the Government Accountability Office (GAO) to report on regulatory impediments to innovation in AI systems. The Protecting Consumers from Deceptive AI Act (H.R. 7766) was introduced on March 21, 2024. The bill would require NIST to establish task forces to support the development of technical standards and guidelines related to distinguishing AI-generated content and providing content provenance data, watermarking, or digital fingerprinting for AI-generated content. The bill would also require a person who makes a software application based on generative AI technology available to users to ensure that the generated content incorporates a disclosure, and implement reasonable measures to prevent such a disclosure from being removed or tampered with. Further, the bill would require providers of covered online platforms to clearly and conspicuously provide such disclosure information. The bill would allow for a person or covered platform to follow self-regulatory guidelines if approved by the FTC.

AI Roundtable Page 189