New Guidance Clarifies Duty to Preserve AI-Generated Legal Data
Legal experts are calling for law firms and companies to preserve AI-generated legal data for litigation.
Why it matters: As AI tools shape legal work, failing to retain AI-generated information could trigger discovery violations. Litigators and in-house counsel must now treat AI-created content like all other key records.
- The National Law Review (May 20, 2026) encouraged proactive preservation of AI-generated data as litigation looms.
- K&L Gates LLP stated that parties must treat relevant generative AI information like any other discoverable ESI.
- A Federal Bar Association CLE instructed attorneys to retain both AI prompts and outputs as part of legal holds.
- Experts stressed that updated protocols should track AI tool usage and maintain clear records of generated content.
Legal teams are facing new clarity around their duty to preserve data created by artificial intelligence as part of pre-litigation planning. A recent update from The National Law Review spells out why generative AI output—such as legal research, memos, or contract language—must be handled like any other discoverable record when a dispute or regulatory event is anticipated.
- K&L Gates LLP, in a February 2026 analysis, confirms: "Relevant GenAI Data is discoverable; and Parties must treat it like any other potentially relevant ESI"—meaning any information, document, or output that can be stored electronically should be preserved if litigation is expected.
- The Federal Bar Association's CLE program in November 2025 specifically addressed retaining AI prompts (the inputs given to generative AI) and outputs, because both qualify as electronic data subject to legal hold requirements during discovery.
- Both commentaries urge firms to create or revise protocols to cover documentation of AI tool usage and ensure that records of all AI-generated content can be preserved and produced in discovery. This includes keeping logs of prompts, results, and any logged interaction that may become relevant if a claim arises.
Legal professionals should act quickly to update e-discovery checklists and notify relevant personnel of obligations to avoid loss or deletion of important AI-related data. As regulatory and case law evolve, expectations around preservation are likely to tighten further, meaning early compliance reduces the risks of evidence loss and court sanctions.
By the numbers:
- May 20, 2026 — Date of National Law Review guidance on AI-generated data retention.
- November 2025 — Date of Federal Bar Association CLE stressing AI prompt and output preservation.
Yes, but: Best practices for preserving some types of AI data remain unclear, as few court rulings directly address AI-generated records.
What's next: Legal teams are watching for the first major court decisions specifically ruling on preservation failures involving generative AI outputs.