Arnold & Porter achieves ~75% cost savings on privilege review with generative AI

While TAR is routinely used for responsiveness review, many attorneys still hesitate to use it for privilege review because it has not proven as effective at identifying privileged documents. Practitioners often default to linear, document-by-document privilege review by human reviewers.

The author, a senior attorney at the firm, used Relativity aiR for Privilege, a generative AI tool built on large language models, to assist with a large document review in an antitrust litigation. A firm attorney spent close to two weeks of focused effort, on a dataset of roughly 500,000 documents, determining whether each law firm, attorney and third party was aligned with the client, adverse, or neutral, so the LLM could assess privilege. aiR assigned each document a privilege prediction, a privilege category, and rationales. The author did not rely on the rationales to a meaningful extent but found the categories very useful for targeting human QC. Humans sampled documents identified as privileged and not privileged; the nuanced “in between” categories required more thorough human review, and a “borderline” category held documents whose privilege status aiR found unclear.

Generative AI streamlined privilege review and gave useful categorization for targeting human review and QC, though it did not eliminate the need for human validation. Relative to the estimated cost of a full linear review with contract attorneys, the author achieved approximately 75% cost savings in this case. The author cautions that other matters may not see similar savings. The post (April 2026) says Relativity charges per document run through the aiR LLM, so aiR may not be cost-effective in matters with large document sets or where significant human review and QC will be necessary, and may not be worthwhile for the smallest document sets given the up-front preparation time.

• • In this matter, a generative AI privilege tool saved approximately 75% against the estimated cost of linear review, but only after substantial up-front attorney effort to determine which law firms, attorneys and third parties confer or break privilege.
• • Relativity markets aiR for Privilege as prioritizing recall over precision, and in the author's experience this bore out: significant human sampling and QC was needed, especially for the nuanced and borderline categories.
• • The author says cost-effectiveness depends on document volume and how much human QC is needed, given the per-document charge for aiR described in the April 2026 post.