AI Litigation Shifts Focus from Training Data to System Behavior
Event summary
- J.S. Held added 42 new AI-related disputes in Q3 2026, bringing total tracked cases to 468.
- 115 AI-related actions have been filed in 2026 through September 30.
- Litigation is expanding from training data issues to post-deployment behavior, including outputs, safeguards, and corporate disclosures.
- Two forms of AI disclosure risk identified: overstating capabilities ('AI washing') and underrepresenting competitive disruption.
- Expert inquiry now requires reconstruction of system operations, controls, and economic impact.
The big picture
The shift in AI litigation from model development to post-deployment behavior reflects broader concerns about system accountability and corporate governance. As courts examine AI outputs, safeguards, and decision-making processes, organizations across the value chain face expanding legal exposure. This trend underscues the need for comprehensive evidence reconstruction capabilities and coordinated expert analysis in AI-related disputes.
What we're watching
- Deployment Risk
- How organizations will demonstrate appropriate controls and system behavior in litigation.
- Disclosure Strategy
- Whether companies can balance AI capability claims with competitive disruption disclosures.
- Expert Coordination
- The pace at which technical, financial, and governance experts integrate evidence collection.
Related topics
