AI Summit_Sept. 13 2024

Congressional Research Service

national security. The focus of the &RPPLVVLRQ¶V work would be to review the federal JRYHUQPHQW¶V current approach to AI oversight and regulation; recommend any governmental structures needed to oversee and regulate AI systems; and build on previous federal efforts and international best practices and efforts to develop a binding risk-based approach to regulation and oversight of AI. The Commission would provide an interim report by six months after appointment of all commissioners, including any proposals for urgent actions, and a final report by six months after the interim report, along with any follow-up reports by one year after the final report. The Commission would terminate 30 days after submitting any follow-up report. The No Section 230 Immunity for AI Act (S. 1993) was introduced on June 14, 2023. The bill would waive immunity under Section 230 of the Communications Act of 1934 for claims and charges against the provider of an interactive computer service if the service used generative AI. 6 The bill would define generative AI as ³DQ [AI] system that is capable of generating novel text, video, images, audio, and other PHGLD EDVHG RQ SURPSWV RU RWKHU IRUPV RI GDWD SURYLGHG E\ D SHUVRQ ´ The Transparent Automated Governance (TAG) Act (S. 1865) was introduced on June 7, 2023. The bill, as reported to the Senate (identical to the TAG Act (H.R. 6886) introduced on December 22, 2024), would direct OMB to issue guidance for federal agencies requiring them to provide disclosure and opportunity for appeal when using certain automated systems and augmented critical decision processes. 7 The head of each federal agency would be required, within 270 days after the date when OMB issues the transparent automated gover QDQFH JXLGDQFH WR LPSOHPHQW LW ³WR WKH H[WHQW WKDW LPSOHPHQWDWLRQ GRHV QRW UHTXLUH UXOHPDNLQJ ´ 7KH *$2 ZRXOG EH UHTXLUHG WR UHYLHZ IHGHUDO DJHQF\ FRPSOLDQFH ZLWK WKH DFW biannually. The bill would sunset after 10 years. The AI Disclosure Act of 2023 (H.R. 3831) was introduced on June 5, 2023. The bill would require that any output of generative $, LQFOXGH WKH IROORZLQJ ³'LVFODLPHU WKLV RXWSXW KDV EHHQ JHQHUDWHG E\ artificial i QWHOOLJHQFH ´ 7KH )7& ZRXOG EH UHTXLUHG WR HQIRUFH WKLV SURYLVLRQ DQG DQ\ SURPXOJDWHG regulations under the section regarding unfair or deceptive acts or practices of the Federal Trade Commission Act (15 U.S.C. §57a(a)(1)(B)). The Artificial Intelligence (AI) Accountability Act (H.R. 3369) was introduced on May 16, 2023. As reported to the House, the bill would require the Assistant Secretary of Commerce for Communications and Information (who is also the Administrator of the National Telecommunications and Information Administration, NTIA) to conduct a study and hold public meetings on accountability measures for AI systems, including how such measures intersect with communications networks and spectrum sharing applications, the promotion of digital inclusion, the reduction of cybersecurity risks, and barriers and challenges to the creation of adequate measures for AI systems. Additionally, the bill would require NTIA to consult with stakeholders through public meetings to develop recommendations about (1) the information that should be available to individuals, communities, and businesses that interact with, are affected by, or study AI systems; and (2) the most effective methods for making that information available. The AI Shield for Kids (ASK) Act (S. 1626) was introduced on May 16, 2023. The bill would require the Federal Communications Commission (FCC), in consultation with the FTC, to issue rules prohibiting entities from offering minor consumers AI features ² including AI chat features ² as part of the products 6 Section 230 of the Communications Act of 1934 generally precludes providers and users of an interactive computer service from being held legally responsible for content provided by a third party. 7 7KH ELOO GHILQHV DQ ³DXWRPDWHG V\VWHP´ DV ³D VHW RI FRPSXWDWLRQDO SURFHVVHV GHULYHG IURP VWDWLVWLFV RU >$,@ WHFKQLTXHV RU WK at otherwise rely on data about specific individuals or groups, to substantially influence the outcome of critical decisions, including computational processes that stand alone or are embedded within another process, system, or application, including paper-based SURFHVVHV´ WKDW ³GRHV QRW LQFOXGH FRPSXWDWLRQDO SURFHVVHV RU LQIUDVWUXFWXUH WKH IXQFWLRQ RI ZKLFK LV QRW GLUHFWO\ UHODWHG WR LQIOXHQFLQJ RU GHWHUPLQLQJ WKH RXWFRPH RI FULWLFDO GHFLVLRQV ´

AI Roundtable Page 193