FTC Gears Up for AI Enforcement: No Brakes in Sight

It is full speed ahead for the Federal Trade Commission’s continuing efforts to understand and determine how to regulate AI. On January 25, 2024, the FTC held its inaugural AI summit.[1] In alignment with the agency’s first algorithmic-bias enforcement action in December and its recent announcement of 6(b) Orders investigating industry’s relationships with and investments in generative AI companies, the summit focused on the threat AI could pose to consumer privacy and the impact of market concentration on competition and innovation. The FTC, eager to avoid what it perceives as its own mistakes concerning corporate consolidations of the 1990s and 2000s,[2] previewed its enforcement priorities for the coming year.

Key Takeaways

• The FTC plans to take an active role in enforcing consumer protection and competition laws against companies using and developing AI technologies.
• The FTC’s focus ranges from chip makers, cloud providers, and AI model-builders to those that use the technology in their products.
• Relief will most likely continue to include the deletion of AI models that allegedly violate the law, including through the use of improperly collected or maintained data.
• Privacy and data security concerns persist with AI software, and the FTC is concerned with potential bias and invasions of privacy.
• Companies that use AI tools should understand how the technology works, what data it processes, and whether third parties (including the AI tool provider) have access to that data, and they must, in certain contexts, be prepared to explain the outcomes of any AI-generated decision-making.

Accelerating Full Stack Enforcement

To start the summit, FTC Chair Lina Khan announced the FTC has opened an inquiry into generative AI partnerships and investments. Foreshadowing potential enforcement actions, the Commission issued 6(b) Orders to three large cloud service providers and two major AI software companies.[3] The move is significant because the FTC is examining partial ownership and partnership arrangements, a sub‑category of mergers recently highlighted in the FTC’s December 2023 Merger Guidelines jointly released with the DOJ Antitrust Division.

The announcement follows the FTC’s November 23, 2023, omnibus resolution authorizing its staff to issue compulsory process requests for non-public investigations involving products and services that use or claim to be produced by AI. The resolution helps to streamline the FTC’s investigations.[4]

Panelist Discussions

The summit consisted of three panels addressing: (1) the role that chips and cloud providers have in AI; (2) the role of data in AI technologies and models; and (3) AI in consumer applications. Panelists articulated concerns about the robustness of competition throughout the AI tech stack and ethical AI deployment. Some of the concerns discussed included:

• AI’s reliance on advanced computing and hardware with limited suppliers;
• Market concentration and its impact on competition and innovation;
• Hurdles faced by startups in entering the semiconductor industry;
• The high costs of AI model training;
• National security concerns about limited chip suppliers;
• The need for transparency in AI applications, particularly in terms of data use and privacy;
• The need for companies to balance innovation with ethical considerations and existing legal and regulatory frameworks to ensure ethical AI deployment; and
• Issues related to bias and data security, as well as the broader societal impacts, such as labor and environmental effects, associated with AI.

No AI Exception to Consumer Protection Laws

According to Commissioner Rebecca Kelly Slaughter, the concentration of AI models with access to consumer data in the hands of a few companies could pose risks around decisions and access to opportunity, information, and privacy. Panelists collectively advocated for the importance of consumer protection in AI, focusing on the implications for privacy and data usage. And both FTC and Consumer Financial Protection Bureau representatives repeatedly noted that “there is no AI exception to the law.”

In his remarks, Commissioner Bedoya highlighted that both old and new AI or algorithmic tools can cause potential violations, such as unfair discrimination.[5] According to him, AI technology success looks like:

• people controlling technology, not the other way around;
• people feeling like they’re in control of technology;
• people knowing where technology is being used to make decisions about them; and
• people knowing why those decisions were made.

What’s Next: Anticipate Scrutiny on AI Marketing, Data Usage, and Retention

The FTC views claims made about AI in marketing materials as a potential enforcement frontier. False advertising claims are of particular interest to the FTC, as such advertisements were characterized by Commissioner Slaughter as both anti-competitive and harmful to consumers. The FTC might also focus on claims around how data is used in connection with AI.

Samuel Levine, director of the FTC’s Bureau of Consumer Protection, closed the day by detailing the bureau’s efforts, including issuing guidance, pursuing enforcement actions, and proposing rules to address AI challenges. He summarized what the public can expect going forward:

• Requirements to delete models trained on illegally collected data;[6]
• Lawsuits against companies that defraud the public with misleading or false claims about AI;
• Rules on new forms of impersonation fraud;
• Limits on retention of children’s data to train models; and
• Enforcement where AI tools potentially harm consumers, including by discrimination, and requirements that companies comply or cease using the tools altogether.

Laura Miller contributed to authoring this client alert.

[1] A transcript of the summit is now available online.

[2] AI Tech Summit, Remarks by Chair Lina Khan, Jan. 25, 2024.

[3] FTC Launches Inquiry into Generative AI Investments and Partnerships, Jan. 25, 2024

[4] FTC Authorizes Compulsory Process for AI-related Products and Services, Nov. 21, 2023

[5] In the past decade, the FTC has repeatedly raised concerns about discrimination generated from data analytics. For example, in 2016, the FTC warned companies that the use of non-representative datasets as inputs may lead algorithms to produce biased results. See Federal Trade Commission, Big Data: A Tool for Inclusion or Exclusion? Understanding the Issues (FTC Report), January 2016, p. 28.

[6] The FTC recently required an AI model to be deleted in its enforcement action against Rite Aid for using racially biased facial recognition security surveillance tools. Prior to Rite Aid, the FTC mandated AI model deletion in several cases related to models or algorithms developed from certain data. See Final Order, In the Matter of Cambridge Analytica, LLC, FTC File No. 18323107 (Dec. 6, 2019); and U.S. v. Kurbo, Inc. et al., FTC Matter No. 1923228 (Mar. 4, 2022).