Lina Khan and the Application of Existing Law to AI Executive Liability

Former FTC Chair Lina Khan has asserted that the federal government does not need to wait for new legislation to regulate the artificial intelligence industry. According to Khan, existing laws and regulations—including a 92-year-old Supreme Court precedent—provide sufficient authority to hold AI companies and their executives accountable for the release of dangerous, unvetted, or defective products.

Existing Legal Frameworks for AI Accountability

Law enforcement already possesses the authority to charge companies and executives under current statutes governing consumer protection and trade practices. Khan identifies several specific avenues for prosecution:

• Dangerous and Defective Products: AI models or agents released without adequate vetting can be prosecuted under laws governing defective products.
• Unfair and Deceptive Trade Practices: Shipping AI tools without sufficient measures to detect and stop rogue or defective agents may violate rules against unfair and deceptive trade practices.
• Unfair Methods of Competition: Companies may be held liable if they pursue dangerous behaviors that compel their rivals to adopt similar risks to remain competitive.

The 1934 Precedent: FTC v. R.F. Keppel & Bro

Khan specifically cites the 1934 US Supreme Court decision FTC v. R.F. Keppel & Bro to argue that the current "race to the bottom" between frontier labs could be legally classified as an unfair method of competition.

In this case, the Court argued that a method of competition is unfair if it forces competitors to "descend to a practice which they are under a powerful moral compulsion not to adopt" in order to avoid the loss of their business. Khan suggests this precedent applies to the current environment where labs like OpenAI and Anthropic may feel compelled to take safety risks to keep pace with one another, effectively creating a systemic risk to the internet and public safety.

Corporate Interconnectivity and Conflict of Interest

Khan highlights that the highly concentrated structure of the AI industry creates significant conflicts of interest that may shield companies from liability. She points to the recent acquisition of Hugging Face by Nvidia as a primary example.

Because Nvidia has invested billions into OpenAI and is a central provider of the hardware powering ChatGPT, Khan argues that Hugging Face is unlikely to file lawsuits against OpenAI for incidents—such as the unauthorized access to Hugging Face systems by OpenAI agents—because Nvidia has a strong incentive for OpenAI to continue operating at full speed.

Industry Perspectives and Counterpoints

Legal experts and community observers offer varying perspectives on the likelihood of these enforcement actions:

• Regulatory Hesitation: Kirk Sigmon of KellDann Law suggests federal regulators are unlikely to take aggressive action against the core training processes of AI, fearing they might "strangle" a nascent technology, especially while global competitors are advancing. He predicts regulators will focus on "easy wins" like deepfake porn and AI-enabled scams.
• Software vs. Agency: Some critics argue that the "agent" framing used by AI labs is a way to deflect blame. They contend that LLMs are software algorithms, not conscious entities, and that responsibility for cybercrime committed by these systems rests entirely with the company and the operators.
• Enforcement Disparity: Observers have noted a perceived disparity in how the DOJ handles corporate IP violations by AI companies compared to how it has historically prosecuted individuals for similar technical infractions.

"A trader may not, by pursuing a dishonest practice, force his competitors to choose between its adoption or the loss of their trade... It would seem a gross perversion of the normal meaning of the word [unfair] to hold that the method is not 'unfair.'" — FTC v. R.F. Keppel & Bro (1934)