By Sarah Reynolds
LaPaglia v. Valve Corp. (S.D. Cal. No. 3:25-cv-00833, filed April 2025)
This case represents the first direct judicial test of whether AI-assisted drafting can undermine an arbitral award’s validity. While still undecided, this case presents important lessons and signals a warning to arbitrators regarding the responsible use of AI. No US court has yet ruled on whether alleged AI authorship alone justifies vacatur under the Federal Arbitration Act.
Background
James LaPaglia, a consumer and small-developer on the Steam gaming platform, initiated an AAA arbitration against Valve Corp., alleging anticompetitive practices and exclusion from Steam’s marketplace. After a consolidated proceeding, the sole arbitrator issued a 29-page final award in Valve’s favor 15 days after final briefing—during the December 23, 2024 – January 7, 2025 holiday period.
The Petition to Vacate
On April 8 2025, LaPaglia petitioned the U.S. District Court for the Southern District of California to vacate the award under FAA § 10(a)(3) and (4), arguing that the arbitrator “outsourced his adjudicative role to Artificial Intelligence.” The filing alleged:
• The award contained factual assertions not in the record (e.g., references to Roblox, Sony–Microsoft cloud partnerships, and Chinese gaming competition).
• The arbitrator had mentioned past use of ChatGPT for writing and a personal desire to complete the award before a trip, suggesting over-reliance on automation.
• A “ChatGPT authorship test” run by petitioner’s clerk produced a model opinion that the award’s prose “suggest[ed] AI generation.”
Valve’s Response and Commentary
Valve denied any improper conduct, emphasizing that no evidence showed the arbitrator actually used AI to decide or draft the award. It argued that stylistic oddities or factual misstatements can occur in human writing and do not meet the FAA’s “exceeded powers” or “misconduct” thresholds.
Key Takeaways for Arbitrators Using AI
1. Maintain Human Control. Drafting tools may assist organization or style, but reasoning, findings, and final language must remain the arbitrator’s own.
2. Disclose Appropriately. If AI tools materially assist in drafting, disclose their use—consistent with emerging guidance from SVAMC and other similar AI protocols.
3. Verify Every Citation and Fact. “Hallucinated” content can constitute factual error or exceedance of authority if left unchecked.
Proactive transparency and diligence are the best defense. Arbitrator’s using AI are encourage to consult the SVAMC Guidelines on the Responsible Use of AI in ADR.
