Calls for rules on artificial intelligence have gained momentum in Washington. At a US Senate hearing, OpenAI CEO Sam Altman and several senators backed international standards, and Altman urged the US to regulate AI and create a new agency. But the debate is not starting from scratch: lawmakers have already proposed measures covering automated decisions, risk assessments, transparency, and data protection.
Earlier proposals offer a starting point
During the Congress that ran from January 2021 to January 2023, politicians introduced a range of AI-related bills. None reached the president’s desk, but policy fellow Anna Lenhart expects some proposals could be revised and return as generative AI tools draw more attention.
Lenhart cautions against treating Congress as if it has no prior experience with AI policy. That framing, she says, can reinforce the technology industry’s preferred argument that lawmakers are too far behind to regulate it. The existing proposals suggest that the policy debate has already touched several practical questions.
Two bills focused on risk and data
Algorithmic Accountability Act
Introduced by Democrats in 2022, before ChatGPT, the Algorithmic Accountability Act addressed harms from automated decision-making. The source article points to systems that denied people pain medications or rejected mortgage applications as examples of the kinds of consequences lawmakers were concerned about.
The bill would require companies to carry out algorithmic impact and risk assessments. It would also put the Federal Trade Commission in charge of regulating and enforcing AI rules, and increase the agency’s staff numbers. Those provisions connect oversight to both examining how systems may affect people and ensuring a regulator has a role in enforcing rules.
American Data Privacy Protection Act
The bipartisan American Data Privacy Protection Act sought to regulate how companies collect and process data. It gained momentum as a way to help women protect personal health data after Roe v. Wade was overturned, but it did not pass in time.
The article says debate over generative AI risks could renew urgency around the proposal. The bill would bar generative AI companies from collecting, processing, or transferring data in a discriminatory way, while giving users more control over how companies use their information. That makes data use one of the existing policy threads that could carry into a broader AI debate.
An agency needs authority to enforce rules
Altman and senators raised the idea of establishing a new US agency to oversee AI. Lenhart argues that the main need is for government to have more technical expertise and resources, whether those sit in a new agency or an existing one that is updated. In either case, the regulator would need enforcement power.
Creating an agency alone would not guarantee effective oversight. Lenhart warns that it is easy to establish a body without giving it powers. Past Democratic efforts to establish new protections through the Digital Platform Commission Act, the Data Protection Act, and the Online Privacy Act failed; the article notes that US bills without bipartisan support are usually unlikely to pass.
The next debate will test bipartisan support
Senators Lindsey Graham, a Republican, and Elizabeth Warren, a Democrat, are working together on a new digital regulator that might also police and perhaps license social media companies. Separately, Democrat Chuck Schumer is organizing Senate support for a bill aimed specifically at AI harms.
Schumer’s effort has bipartisan backing to develop a comprehensive AI bill with safeguards for responsible development. Possible requirements described in the article include letting external experts audit technology before release and providing users and the government with more information about AI systems.
Getting a proposal introduced is only one step. Lenhart says leaders of the commerce committees in both the House and Senate also need to support a comprehensive approach for it to become law. The article argues that lawmakers face a timing challenge: the work needs to advance before public interest in generative AI fades.
The current discussion, then, has several pieces to build on: proposals for assessing automated decisions, privacy protections, and stronger oversight capacity. Whether those ideas become effective rules will depend not just on the design of a new bill or regulator, but on support across Congress and the authority to enforce the laws.