Source: WilmerHale

LinkedIn planned to use aiR, a generative AI-powered document review product within the Relativity platform, to assist in reviewing custodial documents for responsiveness; plaintiffs challenged the planned AI-assisted workflow in Schulte v. LinkedIn Corp., decided in the Northern District of California on June 30, 2026, but the court rejected all of the plaintiffs’ challenges. The court treated generative AI review as a form of traditional Technology Assisted Review (TAR), applying established proportionality, reasonableness, and transparency standards rather than creating a new legal framework for generative AI in discovery. By analyzing the AI workflow under TAR precedents, the court signaled that existing discovery doctrine can accommodate generative AI without wholesale rethinking.

The court specifically upheld LinkedIn’s use of search strings to pre-cull documents before feeding them into aiR because the plaintiffs did not dispute the appropriateness of the search terms themselves; the court relied on TAR cases holding that search terms may be used to reduce review populations prior to TAR. The decision also rejected plaintiffs’ requests for broad AI-related metrics (error rates, elusion estimates, human-validation protocols) absent a demonstrated, particularized deficiency, reiterating that “discovery on discovery” is disfavored. The opinion advises that parties seeking access to AI metrics should consider negotiating particularized ESI-protocol disclosure requirements before discovery begins, and it concludes that thoughtfully designed AI-enabled workflows can meet reasonableness and proportionality standards while delivering modern discovery efficiencies.