Webinar Takeaways: Managing Antitrust Risk When Using Artificial Intelligence — a US Perspective | Epiq

Artificial Intelligence (AI) can help corporations create operational efficiencies, enhance customer experiences, optimize business strategies, significantly reduce costs, and increase profitability. However, using AI tools can create significant antitrust risks for businesses. Those risks include AI’s impact on the accumulation of market power, access of large technology companies to client data, and the relationships between technology companies and AI startups that can circumvent the merger review process.  

In a recent Epiq webinar (co-hosted by the Corporate Counsel Business Journal), a panel of legal experts discussed the antitrust implications of legal and technological developments in AI. During the session, attendees learned about risks inherent in using AI in their businesses and practical steps they can take to ensure their use of AI complies with US antitrust laws.

The panelists included:

Daniel Crane, Professor of Law, University of Michigan

Todd Takashi Itami, Director of Artificial Intelligence and E-discovery Solutions, Covington

Robert Keeling, Partner, Sidley Austin LLP

Moderator: Erin Toomey, Vice President & Managing Director, Epiq Global Investigations Practice Group

TOPIC 1: What AI tools are currently in use and anticipated? What kind of impact will they have?

• Broadly defined, AI is a set of technologies that enable computers to perform a variety of advanced functions, including the ability to see, understand, and translate spoken and written language, analyze data, make recommendations, and more. 

• In the context of antitrust, all the harmful activities possible with computers can also be executed using AI.

• One key distinction: AI models often have a hidden layer with parameters that are not human-readable because of dimensionality, abstraction, and complexity. While it’s possible to run tests, validate, and develop theories by manipulating inputs and observing outputs, peering inside isn’t always feasible.

• The following are high-level buckets of tools in use today: 

  • Foundational Models — General purpose AI like ChatGPT, Gemini, Claude, Llama, Titan, and Stable Diffusion. 

  • Predictive Analytics — AI systems that analyze historical data to make predictions. 

  • Recommendation Systems — AI that suggests content, products, or services based on historical data. 

  • Computer Vision and Audio Processing — AI that processes and interprets visual or audio data. Applications include facial recognition, object detection, and medical image analysis. 

  • Robotics and Autonomous Systems — AI that controls machines and systems capable of performing tasks autonomously. 

• The panel discussed how AI and other technologies will fundamentally change the way the economy works. Notable comments included:

  • “Increasingly what AI is doing is allowing companies to understand consumer demand better than consumers understand themselves.... These technologies will reshape the assumptions about how markets should perform and what markets should do.”

  • “Whatever your information governance policies are for deploying new software, you should follow those. The one distinction is that markets across the board will be rapidly evolving. Developing a culture of technical agility is of paramount importance.”

TOPIC 2: Discussion of antitrust laws applicable to artificial intelligence.

• Section 1 of the Sherman Act and its prohibition on agreements between or among competitors that eliminate or reduce competition.  

• Mergers that eliminate competition by combining companies into a single entity, which also includes certain “partnerships” which may have the impact of a merger. 

• Unilateral exclusionary conduct, which falls under Section 2 of the Sherman Act. Even without combining or conspiring or collaborating with a competitor or another entity, a company that is dominant can suppress competition through predatory conduct. There is a possibility that AI will facilitate predatory pricing schemes. 

TOPIC 3: The antitrust concerns raised by US government agencies and the actions and theories being pursued.

The panel focused its discussion on several recent government investigations and cases:

• The DOJ Section 1 action against RealPage — the complaint alleges that the competitor landlords feed information into RealPage (information exchange) and then RealPage, through its algorithms, makes recommendations about how landlords should price apartments, based upon their layout.  

• Several matters in the US and Europe focus on partnerships and acquisitions. Earlier this year, the FTC issued investigative subpoenas to Alphabet, Amazon, Anthropic, Microsoft, and OpenAI regarding their partnerships, collaborations, and acquisitions of AI-related products.  

• The enforcement activities are focused on the use of AI to collude and anticompetitive agreements or acquisitions involving AI.

• One notable comment: “Companies are going to get smart about not exchanging information or agreeing to use a common algorithm or common AI system. They will unilaterally use advanced AI systems that will have the effect, through iterative machine learning, of giving them the power to raise prices. And I think that the agencies are going to have a really hard time applying antitrust principles to that.

TOPIC 4:  Practical guidance to avoid antitrust exposure when deploying technology and AI tools.

• Follow regulatory guidance — listen to what the regulators are saying because they are speaking publicly with a purpose.
• Conduct diligence to better understand how AI is being deployed in your company. Make sure that legal has a seat at the table.  
• Do the company’s acceptable use policies cover how employees might use AI.
•  Be sure to document the discussions and determinations about how the company is going to   use AI.
• Pay close attention to the products that you are using and what they are designed to achieve.  
• Be vigilant and careful about data sharing — sensitive or competitive information should not be inadvertently shared in ways that could facilitate anticompetitive behavior.
• Make sure that the marketing and PR departments are not over-hyping the capabilities of the products that could draw attention from the regulators. 
• Be particularly attentive to the implications of coordinated conduct when sharing company data.