Algorithmic Accountability Act of 2025
- Sponsor
- Sen. Wyden, Ron [D-OR]
- Committees
- Commerce, Science, and Transportation Committee (primary)
- Last action
- Jun 25, 2025
Bottom line
This bill establishes a comprehensive regulatory framework under the FTC for algorithmic accountability, aiming to mitigate harms from AI systems used in critical consumer decisions by mandating impact assessments and public reporting.
What it actually does
This bill directs the Federal Trade Commission (FTC) to establish regulations requiring certain large entities that deploy automated decision systems or augmented critical decision processes to conduct comprehensive impact assessments. These assessments must evaluate privacy risks, performance, stakeholder consultation, and potential negative impacts on consumers. Covered entities must submit summary reports to the FTC, which will then publish aggregated data and a limited public repository to enhance transparency and accountability.
Proponents argue
Proponents argue the bill is crucial for protecting consumers from biased, discriminatory, or harmful outcomes of increasingly prevalent AI and automated decision systems, ensuring transparency, fairness, and accountability in critical decisions affecting people's lives. It provides necessary oversight in a rapidly evolving technological landscape where algorithms significantly impact access to essential services like housing, employment, and healthcare.
Opponents contend
Potential opponents might argue that the bill imposes significant compliance burdens on businesses, particularly those that might be considered 'covered entities' under the broad definitions, potentially stifling innovation and increasing operational costs. They might also raise concerns about the FTC's capacity to manage such a complex regulatory scheme and the potential for trade secrets to be compromised through reporting requirements, even with limitations.
The bill is moderately detailed and requires careful reading to understand the scope of its definitions and requirements, which is achievable within a reasonable timeframe for those interested in its specifics.
Section 1
Short Title
This section establishes the official short title of the Act, which is the 'Algorithmic Accountability Act of 2025.' This is the common name by which the legislation will be known and referenced in public discourse and legislative tracking.
Supporters argue
A clear short title makes the bill easily identifiable and understandable for public discourse and legislative tracking, aiding in public awareness and legislative efficiency.
Critics contend
No specific opposition is typically raised against short titles, as they are standard legislative practice and serve a practical purpose.
Tradeoffs
None.
Section 2
Definitions
This section meticulously defines the core terminology used throughout the Act. Key definitions include 'automated decision system' (any AI/ML system used for decisions), 'augmented critical decision process' (a process using an automated decision system for a critical decision), 'critical decision' (decisions with significant effects on a consumer's life, such as those related to education, employment, housing, financial services, healthcare, and more), and 'covered entity' (companies meeting certain revenue/equity thresholds or data usage volumes, deploying such systems).
Section 3(b)
FTC Regulations for Impact Assessments
This section mandates the Federal Trade Commission (FTC) to issue regulations within two years of the Act's enactment. These regulations will require 'covered entities' to conduct ongoing impact assessments of their automated decision systems and augmented critical decision processes, both before and after deployment. Entities must maintain documentation of these assessments and submit annual summary reports to the FTC, with an initial report required for new systems. The regulations also require meaningful consultation with stakeholders and timely mitigation of likely material negative impacts on consumers.
Section 4(a)
Impact Assessment Requirements
This section outlines the specific requirements for the impact assessments that covered entities must perform. These include evaluating previously existing decision processes, identifying and describing stakeholder consultations, conducting ongoing testing for privacy risks and performance (including differential performance across demographic groups), training employees, assessing limitations on system use, maintaining documentation of data used, evaluating consumer rights (notice, opt-out, contestation), identifying and mitigating negative impacts, and documenting the development and deployment process. It also allows for documenting infeasibility with rationale.
Section 5
Summary Report Content
This section details the information that covered entities must include in the summary reports submitted to the FTC. These reports must draw from the full impact assessments and include details such as the covered entity's identity, a description of the critical decision and system's purpose, stakeholder consultation information, testing and evaluation results (including differential performance), publicly stated limitations on system use, data sourcing details, transparency measures, identified negative impacts and mitigation steps, and a list of any assessment requirements that were infeasible to comply with, along with the rationale.
Section 6
FTC Reporting and Public Repository
This section requires the FTC to publish an annual report summarizing aggregated information from the submitted summary reports, focusing on broad trends and anonymized lessons learned. Crucially, it also directs the FTC to establish a publicly accessible repository within 180 days of regulations taking effect. This repository will contain a limited subset of information from the summary reports, such as the covered entity's identity, the critical decision category, publicly stated prohibitions on system use, data sources, performance metrics, and links for consumer recourse, to inform consumers and researchers and ensure compliance.
Section 8(a)
Establishment of FTC Bureau of Technology
This section establishes a new Bureau of Technology within the Federal Trade Commission, to be headed by a Chief Technologist. The Bureau's duties include advising the Commission on technological aspects of its functions, aiding in the enforcement of this Act, and providing technical assistance for investigations. It authorizes the Chair to appoint at least 50 personnel with expertise in fields like technology, digital design, cybersecurity, civil rights, and machine learning within two years, outside of civil service laws.
Section 9
Enforcement Mechanisms
This section stipulates that violations of the Act or its regulations will be treated as unfair or deceptive acts or practices under the Federal Trade Commission Act, granting the FTC full enforcement powers, including penalties, as if the provisions were part of the FTC Act. It also authorizes State attorneys general to bring civil actions on behalf of their residents, with requirements for notifying the FTC and allowing for FTC intervention, creating a dual enforcement model.
The definition of a 'covered entity' includes specific financial thresholds (e.g., >$50M in average annual gross receipts or >$250M in equity value) or data usage thresholds (using identifying information about >1,000,000 consumers, households, or devices). It also includes entities that met these criteria within the previous 3 years and incorporates an inflation adjustment.
Section 2(7)
Why it matters:It is standard legislative practice to define the scope of applicability precisely, but the complexity of the thresholds can obscure the full reach of the bill, potentially making it harder for businesses to quickly determine if they are covered.
Case for: Supporters argue these thresholds are designed to capture large entities with significant impact on consumers while exempting smaller businesses from potentially burdensome regulations. The look-back period ensures entities cannot easily evade compliance by restructuring or temporarily falling below thresholds.
Case against: Critics might argue the thresholds are still too low, capturing many medium-sized businesses, or that the data usage threshold (1,000,000 consumers) is easily met by many online services, leading to widespread applicability and significant compliance costs for a broad range of companies.
Estimated impact: Significantly impacts a large number of technology companies, financial institutions, healthcare providers, and other businesses that use AI in critical decision-making, potentially requiring substantial investment in compliance infrastructure and personnel.
This clause allows the Commission to determine, through rulemaking, 'any other service, program, or opportunity decisions about which have a comparably legal, material, or similarly significant effect on a consumer's life' as a critical decision.
Section 2(8)(I)
Why it matters:This is standard legislative practice to provide regulatory agencies with flexibility to adapt to future circumstances and technologies that cannot be foreseen at the time of enactment. However, it also grants significant power to the FTC to expand the bill's scope without further congressional action.
Case for: Proponents argue this flexibility is essential to future-proof the legislation, allowing the FTC to address new critical applications of AI that emerge after the bill's enactment without needing new legislation, ensuring ongoing consumer protection.
Case against: Opponents might argue this grants excessive power to an unelected agency, creating regulatory uncertainty for businesses who may not know if their systems will become subject to regulation in the future, potentially chilling innovation in new areas.
Estimated impact: Potentially broad expansion of the bill's scope over time, affecting new industries and applications of AI as technology evolves and the FTC exercises its rulemaking authority.
This provision states that it shall be unlawful for any covered entity to commit the prohibited acts (violating FTC regulations) 'regardless of specific agreements between entities or consumers.'
Section 3(a)(2)
Why it matters:This clause is a technical clarification intended to prevent companies from using private contracts or terms of service to circumvent the consumer protections mandated by the Act. While technically phrased, it has significant implications for contractual relationships.
Case for: Supporters argue this ensures that companies cannot use private contracts or terms of service to circumvent the consumer protections mandated by the Act, ensuring universal applicability and preventing loopholes that could undermine the bill's intent.
Case against: Critics might argue this provision interferes with freedom of contract and could invalidate existing agreements, creating legal challenges and uncertainty for businesses that have structured their relationships based on private terms.
Estimated impact: Ensures broad applicability of the Act's protections, preventing companies from contracting out of their responsibilities and establishing a baseline of consumer rights that cannot be waived by agreement.
This subsection clarifies that the Act does not require a covered entity to share with or otherwise disclose to the Commission or the public any information contained in a full impact assessment performed in accordance with this Act, except for the information required in the summary report.
Section 4(c)
Why it matters:This is a crucial carve-out for businesses, balancing the need for transparency with the protection of proprietary information. It is designed to reduce industry opposition by limiting the scope of mandatory public disclosure, which might be less emphasized than the disclosure requirements themselves.
Case for: Proponents argue this provision protects proprietary information and trade secrets of businesses, encouraging compliance with the assessment requirements by reducing the risk of competitive disadvantage from full public disclosure. It strikes a balance between transparency and commercial interests.
Case against: Critics might argue this limits transparency and public oversight, making it harder for external parties, including researchers and advocates, to fully scrutinize the assessments and verify their thoroughness and objectivity, potentially hindering accountability.
Estimated impact: Provides significant protection for business proprietary information, potentially reducing industry opposition to the bill's assessment requirements and facilitating smoother implementation, while still allowing for some public insight through summary reports.