Berger v. Graf Acq., LLC, C.A. No. 2023-0873-LWW (Del. Ch. Oct. 21, 2024) (Will, V.C.)

In this letter opinion, Vice Chancellor Will (the “Court”) held, among other things, that a producing party could use technology assisted review (“TAR,” also known as predictive coding) in connection with responding to document requests in discovery.

The case involved putative class action claims brought by a former stockholder of Graf Industrial Corp. against the company’s former directors and officers.  During discovery, the plaintiff requested documents using a search query that would have required the defendants to review 125,000 documents.  The defendants rejected the proposal, arguing that the request was unduly burdensome and disproportionate to the needs of the case.  The defendants proposed that the plaintiff either narrow the search terms or permit defendants to use TAR to assist with the review, which, unlike a manual review of each document by a human, leverages machine learning techniques with the guidance of attorneys.  As the Court explained, attorneys train the TAR system through iterative rounds of sampling and manual review, beginning with the coding of documents in a representative seed set.  Once the TAR system uses the training to make relevance determinations across the broader set of documents, attorneys are responsible for quality control, including by reviewing and running keyword searches over documents coded by the program as non-responsive.

The plaintiff rejected these proposals and filed a motion to compel, insisting that the defendants conduct a manual review of the 125,000 documents that hit on the proposed search terms.  In granting the plaintiff’s motion in part and denying it in part, the Court ruled that the defendants should apply the broader search terms but that they could leverage TAR in conducting the review.  The Court explained that a producing party could use TAR in e-discovery as long as the producing party is transparent with the court and opposing counsel with respect to the process employed.  As the Court explained, “[i]t is not up to the requesting party to block TAR . . . . [n]or is it necessarily a matter for the court to dictate,” because producing parties are in the best position to decide how to review their own electronically stored information (“ESI”).  The Court also noted that TAR, in addition to promoting efficiency, has been shown to be at least as accurate as manual review.

This opinion coincided with the release of the Delaware Supreme Court’s new guidance on how judicial officers and court personnel should use generative artificial intelligence (“GenAI”).  Link to press release.  Link to order.  The interim policy states that users should remain responsible for content they produce, input non-public information only into GenAI tools approved by the state, and not delegate decision making to GenAI.

About Potter Anderson

Potter Anderson & Corroon LLP is one of the largest and most highly regarded Delaware law firms, providing legal services to regional, national, and international clients. With more than 100 attorneys, the firm’s practice is centered on corporate law, corporate litigation, intellectual property, commercial litigation, bankruptcy, labor and employment, and real estate.