Plaintiffs firm wins motion after AI audit exposes defense overproduction of 3.5M documents

In litigation involving major commercial airlines, the defendant produced 3.5 million documents under a court deadline using Technology Assisted Review (TAR). The defense TAR Control Set estimated the production would contain a substantial majority of responsive documents at 58% precision (about a 1:1 responsive-to-non-responsive ratio). Plaintiffs needed to verify whether that production was accurate before continuing review and case strategy.

Mid-case, the leading plaintiffs co-counsel reanalyzed the defense production in Casepoint’s legal discovery platform using CaseAssist active learning and advanced analytics. They took a Validation Sample and applied TAR 2.0 features to audit the data and compare Casepoint’s findings against the defendant’s original TAR 1.0 Control Set estimates.

The audit found the production actually contained nearly all responsive documents, but precision was only 16.7%—shifting the review ratio from 1:1 to 4:1 and supporting an overproduction claim. Casepoint’s AI reduced the 3.5 million produced documents down to 600,000 responsive documents. Defendants confirmed an error in their control-set estimates, and the court granted plaintiffs’ motion to extend.

• Built-in TAR 2.0 / active learning can independently validate an opposing party’s eDiscovery production estimates mid-case
• Large precision gaps between claimed and audited TAR results can substantiate overproduction and support successful motions
• Reanalyzing a high-volume production with platform AI can quickly cull non-responsive volume and accelerate case strategy