What happened
Five women sued Publicis Groupe and its US subsidiary MSL Group for gender discrimination in the US District Court for the Southern District of New York. MSL proposed to cull about three million emails with predictive coding, and the plaintiffs accepted its use in principle. Its protocol named the Axcelerate software by Recommind, trained on seed sets that attorneys code, in up to seven rounds. On 24 February 2012 Magistrate Judge Andrew J. Peck approved the method. He weighed five factors, among them the volume of over three million documents and the superiority of computer-assisted review to manual review or keyword searches. He wrote that the opinion recognizes computer-assisted review as an acceptable way to search for relevant electronically stored information in appropriate cases. District Judge Andrew L. Carter Jr. adopted the rulings in April 2012. The superiority rests on published studies. The review had not run, and the opinion reports no result from it.
Sources
-
Court document, by authors with no tie to the system.
United States District Court for the Southern District of New York (Magistrate Judge Andrew J. Peck). Moore v. Publicis Groupe & MSL Group, 287 F.R.D. 182 (S.D.N.Y. 2012), Opinion and Order, 24 February 2012. Caselaw Access Project copy of the Federal Rules Decisions report.
In this case, the Court determined that the use of predictive coding was appropriate considering: (1) the parties’ agreement, (2) the vast amount of ESI to be reviewed (over three million documents), (3) the superiority of computer-assisted review to the available alternatives (i.e., linear manual review or keyword searches)
Details
AI system
- Workflow
-
The parties agreed to use predictive coding and disagreed on parts of the protocol. Attorneys reviewed a random sample of 2,399 documents to build the seed set that trains the software. The protocol allowed up to seven training rounds and shared the coding with the plaintiffs' counsel.
- Relation to AI
-
The court approved the review before it ran. The protocol describes software trained on seed sets that attorneys code.
Follow-up
- Conditions for replication
-
The document collection is large, here over three million emails. The producing party tests quality and keeps its process transparent to the other side. The cost of review stays proportionate to what is at stake in the case.
Entry details
- Sector
-
Legal services / e-discovery.
- Purpose
-
Recommind built the Axcelerate software to search and review large document collections, with predictive coding that learns from documents that attorneys code.
Not yet checked: model, release, provider, role, configuration, inputs and outputs, autonomy, human oversight, AI contribution, beneficiaries, reach, distribution and access, duration, every measurement question, every money question, ownership and control, dependencies, cross-border connections, geopolitical implications, replication attempts, availability, institutional decision, grounds, action and result, success criterion, every evidence question, observed or intended, alternative assessments, related entries, other identifiers, category mappings.
Cite this record
Peter H. Vartanian. (2026). Federal court approves Axcelerate for three million emails (AIBD-0828). Artificial Intelligence Benefits Database, edition of 31 October 2026. https://petervartanian.xyz/aibd/records/AIBD-0828.html (archived at https://doi.org/10.7910/DVN/AADSII, version 4.0)