H.R. 5511: Algorithmic Accountability Act of 2025

H.R. 5511 would create a new federal accountability system for algorithms used in what the bill calls “critical decisions” — including education, employment, utilities, family planning and reproductive services, mortgages and credit, health care, housing, lodging, and legal services. A “covered algorithm” is defined broadly to include machine learning, natural language processing, AI, and similar computational processes that create or facilitate products, rank information, make decisions, or help humans make decisions in these areas.

The bill is aimed mainly at larger firms. Tier 1 covered entities include companies with more than $50 million in average annual gross receipts or more than $250 million in equity value over a 3-taxable-year period, or companies that manage identifying information for more than 1,000,000 consumers, households, or devices. Tier 2 reaches smaller firms — those with more than $5 million in average annual gross receipts or more than $25 million in equity value over a 3-taxable-year period — if they deploy algorithms intended for use by Tier 1 entities. Companies remain covered if they met the threshold at any point in the previous 3 years, and those dollar thresholds rise annually with the Consumer Price Index after the first fiscal year after enactment.

The core requirement is impact assessment. The FTC must write regulations within 2 years of enactment, and those rules take effect 2 years after they are promulgated. Covered entities would have to assess algorithms both before deployment and after deployment, keep documentation for 3 years longer than the algorithm stays in use, and file annual summary reports for ongoing systems plus initial summary reports before new systems launch. Those assessments must look at privacy and security, including data minimization and tools like federated learning and differential privacy, and test whether systems perform differently based on race, color, sex, gender, age, disability, religion, family status, socioeconomic status, or veteran status. Companies also must evaluate notice, opt-out options, transparency, explainability, and whether people can contest or appeal decisions.

The bill also builds an enforcement and transparency structure around those reviews. The FTC would have to create a public repository within 180 days of issuing regulations, make it publicly accessible within 180 days of the rules taking effect, and update it quarterly with searchable, sortable, downloadable summary reports. Starting 1 year after the effective date, the FTC would publish annual trend reports with aggregated statistics. To do this, the bill creates a Bureau of Technology at the FTC led by a Chief Technologist, requires at least 50 personnel to be appointed within 2 years of enactment, and allows 25 additional enforcement staff in the Bureau of Consumer Protection. Violations would be treated as unfair or deceptive acts or practices under the FTC Act, and state attorneys general could also sue after giving written notice to the FTC — or immediately upon filing if advance notice is not feasible. The bill does not preempt state, tribal, city, or local laws, so it would add a federal floor without wiping out stricter local rules.