How AI Assists With Antitrust Document Review in Merger Investigations

When the FTC or DOJ issues a second request in a merger investigation, the responding party faces one of the most demanding document review exercises in legal practice. The scope is enormous, the deadline is firm, and the stakes are a deal worth billions of dollars. Missing a key document can mean extended investigations, consent decrees, or blocked transactions.

AI-assisted document review has become standard practice in this context, and for good reason. The volume of material is simply too large for traditional linear review to handle effectively within the typical timeframe.

The Scale of Second Request Review

A second request from a federal antitrust agency typically requires producing documents related to the competitive dynamics of the markets affected by the merger. For large companies, this means searching across email systems, document management platforms, shared drives, messaging applications, and sometimes personal devices of key custodians.

The initial collection can easily reach tens of millions of documents. After de-duplication and filtering, you might still be looking at several million documents that need some level of review. Traditional review teams of contract attorneys working through documents one by one simply cannot process this volume within a reasonable timeframe while maintaining quality.

How AI Transforms the Review Process

Predictive coding and continuous active learning. AI-powered review platforms use machine learning to prioritize documents most likely to be responsive. The system learns from attorney coding decisions on a seed set of documents and applies those patterns to rank the remaining population. As attorneys continue reviewing, the system continuously updates its model, becoming more accurate over time.

This approach means that the most important documents surface early in the review, giving the deal team visibility into potential issues before the full review is complete. It also means that large portions of the document population that are clearly non-responsive can be identified and excluded with statistical confidence, dramatically reducing the number of documents requiring human review.

Concept clustering and visualization. AI groups documents by topic and concept, allowing attorneys to see the structure of the document population before diving into individual documents. In a merger context, this might reveal clusters around competitive pricing discussions, market share analysis, customer overlap, or integration planning. These clusters help the review team allocate resources to the most critical areas first.

Key document identification. Antitrust agencies are particularly interested in certain types of documents: strategic plans, competitive analyses, board presentations about the merger rationale, and communications about pricing or market allocation. AI can be trained to identify these document types specifically, ensuring they receive priority review and are flagged for senior attorney attention.

Privilege Review at Scale

Privilege review in a second request response is its own challenge. Attorney-client privileged communications and work product need to be identified and withheld or redacted before production. With millions of documents in the review population, the privilege review needs to be both thorough and efficient.

AI assists by identifying documents that contain indicia of privilege: communications with or copying attorneys, references to legal advice, draft documents with attorney comments, and communications with outside counsel. The system can also track privilege threads, ensuring that if one email in a chain is privileged, the system flags the entire thread for review.

For firms handling the privilege log, AI can extract the metadata needed for each log entry and generate draft privilege descriptions. This automation is valuable because privilege logs in second request responses can contain thousands of entries, and manually compiling each entry is extremely time-consuming.

Hot Document Escalation

In merger investigations, certain documents can make or break the deal. A single email from a senior executive describing the merger as a way to eliminate a competitor can shift the entire agency analysis. AI systems can be configured to flag documents containing language indicative of anticompetitive intent, market allocation discussions, or pricing coordination.

These hot documents are escalated immediately to senior attorneys and the deal team, even before the broader review is complete. Early identification of hot documents allows counsel to prepare explanations and context before the agency sees them, rather than being caught off guard during the investigation.

Production Management

Beyond the review itself, AI helps manage the production process. Automated redaction of sensitive information not relevant to the investigation, consistent application of confidentiality designations, and tracking of production volumes against compliance deadlines are all areas where AI reduces the administrative burden on the review team.

For rolling productions, which are typical in second request responses, AI can optimize the order in which documents are produced to satisfy agency requests for priority categories while maintaining an efficient overall review workflow.

Quality Control

Antitrust regulators take production completeness seriously, and inadvertent failures to produce responsive documents can create significant problems. AI-powered quality control measures include statistical sampling to validate coding accuracy, identification of documents that the model finds difficult to classify for additional human review, and comparison of coding decisions across reviewers to identify inconsistencies.

These quality measures provide a defensible record of the review methodology, which is important if the agency later questions the completeness of the production.

The Competitive Reality

At this point, AI-assisted review is not a competitive advantage in antitrust merger work. It is the baseline expectation. Firms that do not use these tools are at a significant disadvantage in terms of both cost and quality. The firms that are differentiating themselves are those that have refined their AI workflows through experience on multiple second request engagements, achieving better results with fewer review hours.

If your firm handles merger review work, AI document review is a necessary investment. For a broader look at AI applications in legal practice, see FirmAdapt's law firm solutions page.