The U.S. is generally viewed as “behind” in its regulation of AI compared to the European Union and Asian countries. Yet ChatGPT’s release triggered a tsunami of U.S. legislation in 2023 from federal and state legislators seeking to address perceived concerns with the emerging and fast evolving technology. State legislatures have introduced nearly 200 AI bills in 2023. Congress does not have nearly that number of AI bills, with about 30 bills tabled thus far. The various pieces of U.S. legislation – federal or state – seek to regulate both the creation of AI models and how those models may be used.
In gridlocked Washington, D.C., a comprehensive federal approach to AI – including the U.S. Government’s role in regulating it – has received bipartisan support. This is likely because AI regulation overlaps nicely with bipartisan issues like national security concerns vis-à-vis the People’s Republic of China; transparency issues (e.g., uses of the technology, and the existence of a human decisionmaker component in “higher risk” AI use categories); deep fakes and election/disinformation concerns; safeguarding against the exploitation of children; and licensing/ensuring copyright protection of existing works. Of course, partisan approaches to AI policymaking also exist. Republican lawmakers want guardrails that do not stifle American innovation, recognizing that AI has a role in ensuring America’s competitive edge globally, while democratic lawmakers tend to focus on the prevention of bad impacts on consumers and workers based on the potential (mis)use of AI.
For example, on September 21, Senator Ron Wyden (D-Oregon) introduced the Algorithmic Accountability Act of 2023 (S. 2892), a bill with 11 other Democratic Senators. The bill would direct the Federal Trade Commission (FTC) to require impact assessments of automated decision systems and augmented critical decision processes. The House has a companion (identical) measure (H.R. 5628) with only Democratic sponsors. Democratic lawmakers have also introduced legislation focused on disclosing whether output has been generated by AI; prohibiting or requiring disclosure of, the surveillance, monitoring, and collection of certain worker data by employers via AI systems; and prohibiting employers from relying exclusively on an automated decision system in making an employment-related decisions.
Republicans have introduced fewer GOP-only bills in Congress. One GOP-only bill would promote the use of AI – H.R. 206, the Healthy Technology Act of 2023. This legislation would establish that AI or machine learning technology may be eligible to prescribe drugs by amending the Federal Food, Drug, and Cosmetic Act. Given the divided Congress, bills that have bipartisan sponsors are better positioned to advance to the President’s desk.
The Senate is moving forward with a bipartisan effort to devise and advance AI legislation. Senate Majority Leader Charles Schumer (D-New York) and his designated sherpas – Senators Martin Heinrich (D-New Mexico), Todd Young (R-Indiana), and Michael Rounds (R-South Dakota) – are leading the effort. Majority Leader Schumer has indicated a preference for a committee-by-committee approach to AI regulation, which will slow the legislation down but also move the Senate toward a more comprehensive plan.
Majority Leader Schumer is also hosting AI briefings to help Senators better understand AI before legislating, which also takes time. Some Senators are ready to move forward with AI legislation; others seek to listen-and-learn before legislating. Notably, Senator Maria Cantwell (D-Washington), Chair of the Senate Commerce Committee, believes Congress should already have a decent idea of what it would like to accomplish on AI. She recently reminded lawmakers: “I set up the [National AI Advisory Committee] years ago for this very thing: what should the government role be?” According to Senator Murray, Congress has already had “three years” to discuss what AI legislation should “really look like” and she is ready to move ahead. Senators Richard Blumenthal (D-Connecticut) and Josh Hawley (R-Missouri) are likewise seeking to move AI legislation forward at a faster pace. In September, the Democratic chair and top Republican on the Senate Judiciary Subcommittee on Privacy, Technology and the Law outlined legislation that would create an independent oversight agency for artificial intelligence technologies and require companies creating high-risk applications to register with the new body.
Lawmakers are also facing some serious logistical constraints – namely, a busy legislative calendar for the remainder of this year leaving little time to legislate comprehensively on AI, which has been further complicated with the ongoing selection of a new Speaker of the House; the 2024 elections next year that tends to make the environment harder to advance legislation; and a new Congress (119th) that will be sworn-in in January 2025, starting the clock over for any legislation that fails to be approved during the 118th Congress. Some therefore believe comprehensive federal legislation is not on the horizon this year, and instead smaller (i.e., not comprehensive) bills could possibly advance during the remainder of the current Congress (118th).
Even if lawmakers in Washington agree on addressing issues and concerns, there are frequently partisan disagreements over how best to do so. This problem has dogged efforts to pass federal data privacy and, more recently, cryptocurrency legislation. It could continue to delay any comprehensive AI legislative package.
As the largest employer in the U.S., lawmakers may instead shift and continue to push targeted bills that focus on the U.S. Government and its use of AI. A narrower approach that has quietly gained momentum is spearheaded by Senator Gary Peters (D-Michigan). His bill (S. 1865), the Transparent Automated Governance (TAG) Act, would regulate AI through the federal government’s use and procurement of the technology. The bill has already been marked up and approved at the committee level. Senator Peters, who chairs the Senate Committee on Homeland Security & Governmental Affairs, has a proven track record of sponsoring bills with Republican lawmakers that that have become law, such as (S. 2551), the Artificial Intelligence Training for the Acquisition Workforce Act (“AI Training Act”) in the 117th Congress. Senator Peters has other bipartisan AI bills tabled for consideration in the current (118th) Congress – a measure that would create training programs on AI for federal officials (S. 1564), and one that would mandate federal agencies designate a point person for AI (S. 2293).
Separately, the Biden Administration has and continues to announce executive branch-led actions to address AI, with an AI Executive Order due out this Fall. Meanwhile, some U.S. states are moving forward with AI legislation in the absence of a federal legislative policy, creating a patchwork of state laws on the emerging technology.
Navigating the evolving Federal Executive Branch vs. Federal Legislative vs. State Legislative approaches to AI requires considerable diligence. That the legislative process – federal and state – is ongoing now presents a unique opportunity for AI technology organizations to shape AI policy, such as advocacy to congressional delegation through educational outreach and awareness of the benefits. We highly recommend that businesses engage early in shaping AI policy, rather than wait on the sidelines for final rules, laws, or resulting litigation. Those that do not engage will have to continually adapt their due diligence structures to the federal and/or other policies that eventually emerge.