The Limits of Disclosure: Artificial Intelligence, Arbitration, and the Illusion of Informed Consent

Authors

• Rishi Batra

Abstract

Consider the following scenario: A warehouse worker brings a wage and hour claim against her employer under her state's labor code. Her employment contract, which she signed on her first day of work, contains a mandatory arbitration clause that she did not negotiate, did not fully read, and — like most employees in her situation — did not fully understand. An arbitrator is appointed, a hearing is held, and an award is issued in the employer's favor several weeks later, denying the employee's claims entirely. The award is well-organized and internally consistent, addressing each of the worker's claims with what appears to be careful legal reasoning. What the worker does not know — and was never told — is that the arbitrator used a large language model to draft the substantive findings and legal analysis, reviewing the AI-generated output only briefly before signing. The award is final and binding. Under the Federal Arbitration Act, the grounds for vacatur are narrow, and none of them — at least on their face — obviously reaches the AI use or the fact that the arbitrator did not disclose it. The worker has no idea that the decision that determined whether her wages were lawfully withheld was substantially generated by an algorithm trained on millions of documents, with biases she cannot identify and limitations she cannot challenge.

Recommended Citation

Rishi Batra, The Limits of Disclosure: Artificial Intelligence, Arbitration, and the Illusion of Informed Consent, 2026 J. Disp. Resol. ()
Available at: https://scholarship.law.missouri.edu/jdr/vol2026/iss2/6