Responding to an FTC Civil Investigative Demand Targeting AI and Machine Learning

Responding to an FTC Civil Investigative Demand Targeting AI and Machine Learning

Artificial intelligence (AI) and machine learning have become integral parts of many businesses and industries. From chatbots to recommendation engines to fraud detection, AI and ML are driving innovation and efficiency across sectors. However, the rapid pace of development has also raised regulatory concerns about potential harms, biases, and lack of transparency.

In recent years, the Federal Trade Commission (FTC) has taken greater interest in emerging technologies like AI and ML. Using its authority under Section 5 of the FTC Act to prohibit “unfair or deceptive acts or practices,” the agency has begun investigating whether certain uses of AI/ML could violate consumer protection laws. This includes issuing Civil Investigative Demands (CIDs) to companies requiring them to provide information and documents related to their development and use of AI/ML systems.

Common Areas of FTC Concern with AI/ML

Based on recent CIDs and public statements, several key areas of AI/ML attracting FTC scrutiny include:

  • Bias and fairness – Whether algorithmic systems discriminate against protected groups
  • Transparency – Whether companies are clear with consumers about their use of AI/ML
  • Data privacy – How consumer data is collected/used to train AI models
  • Accuracy – Whether AI systems generate consistent, accurate results

For example, the FTC questioned companies using AI for employment decisions about the steps taken to prevent discrimination. The agency has also suggested that overly broad claims about a product’s “AI” capabilities could be deceptive to consumers under Section 5.

Responding to an FTC CID

Receiving a CID related to AI/ML can be intimidating, but companies have certain rights in the investigatory process. Here are some tips if your organization gets an AI/ML-focused CID from the FTC:

  1. Carefully review the CID – Make sure you understand the exact information and materials being requested
  2. Meet with counsel – Experienced attorneys can help craft your response strategy. Consider FTC defense lawyers who have worked on prior AI cases
  3. Negotiate the scope – Try narrowing overly broad/burdensome CID requests that extend beyond AI issues
  4. Collect responsive documents – Gather materials from departments involved in AI/ML development, testing, and implementation
  5. Craft explanatory responses – Provide context around development processes, data usage, and other areas of interest
  6. Consider proactive changes – The CID may reveal areas to improve processes before violations occur
Phase Key Actions
Initial Review Assess CID scope, form response team, retain legal counsel
Meet with Counsel Discuss compliance strategy, grounds for narrowing demands
Collect Documents Gather materials from AI/ML teams and other stakeholders
Written Responses Draft narrative responses explaining processes, contexts
Production Produce documents and responses to FTC by deadline

The FTC provides guidance on responding to CIDs, but working with experienced counsel is highly recommended when AI/ML practices are the focus. Even if no actual violation occurred, the FTC may pressure companies to alter AI development processes deemed potentially unfair or deceptive.

AI/ML Compliance Takeaways

The likelihood of additional FTC investigations into AI/ML means companies should proactively assess compliance risks. Smart strategies include:

  • Cataloging all AI/ML systems and their core functionality
  • Establishing rigorous processes for testing systems before deployment
  • Creating explainability reports on high-risk AI models
  • Assembling a cross-disciplinary team to oversee AI/ML risks
  • Regularly reviewing AI practices against latest regulatory guidance

Documenting efforts like these can help demonstrate good faith compliance in an FTC investigation. And identifying areas for improvement may prevent violations altogether.

The regulatory framework around AI/ML remains fluid and complex. But prioritizing transparency, fairness, and accountability provides the best insulation against legal risk. Companies taking proactive steps today will be best positioned for the more clearly defined regulatory landscape of tomorrow.