Rabiej Webinar (3 of 3) – Barriers to AI in eDiscovery

Barriers to AI in eDiscovery

Electronic discovery in mass-tort multidistrict litigations (MDLs) has ballooned in scale, with nearly 190,000 cases pending in federal courts as of 2025, predominantly product liability disputes. These proceedings consolidate thousands of similar claims, generating vast volumes of electronically stored information—often millions of documents per case—that must be preserved, collected, reviewed, and produced under tight timelines.

Recent years have seen generative AI tools promise dramatic reductions in review time and cost, with some providers bundling AI features into standard pricing and others reporting up to 62% adoption for document review automation. Yet industry reports from 2025 and early 2026 highlight an 'AI gap': adoption lags behind capability due to skepticism about reliability, fears that AI outputs lack defensibility in court, and worries over data privacy when tools process sensitive or confidential information.

The stakes are concrete. In MDLs, overbroad discovery practices lead to productions of ten million documents or more, driving costs into the tens of millions of dollars and straining resources for both plaintiffs' leadership counsel and corporate defendants. Courts, through amendments to Federal Rule of Civil Procedure 26 emphasizing proportionality, have pushed back against excess, but without effective AI tools, parties struggle to comply efficiently. Inaction risks sanctions, missed deadlines in coordinated pretrial schedules, or weakened positions when critical evidence remains buried.

Non-obvious tensions emerge between stakeholders. Defense teams often control advanced technology-assisted review (TAR) systems, while plaintiffs face resource constraints in managing incoming productions. Meanwhile, courts balance encouraging innovation against risks of bias, hallucinations in generative outputs, or privacy breaches—evident in 2025 rulings compelling preservation of AI training logs and output data in copyright infringement MDLs, even as anonymization mitigates some concerns. Smaller firms and plaintiffs' counsel particularly face barriers from high initial costs and the need for specialized expertise, potentially widening access gaps in high-stakes litigation.

As 2026 begins, the combination of falling AI tool prices, judicial familiarity with advanced search methods, and growing case law on AI-related discovery obligations intensifies pressure to resolve these barriers.