- Five named firms building custom AI: Kirkland, Clifford Chance, Crowell & Moring, Freshfields, Fried Frank.
- Polsinelli is the named outlier still using third-party tools (Harvey, Legora).
- No reporting discusses privilege, confidentiality, or vendor-diligence implications of feeding client and matter data into proprietary AI systems.
- For solo/small/mid-market firms, the right takeaway is that internal-data AI is not the only path to AI advantage. Operational discipline around off-the-shelf tools remains the affordable wedge.
BigLaw firms compete on internal-data AI: Kirkland, Clifford Chance, Freshfields, Crowell, Fried Frank.
A wave of large global firms (Kirkland, Clifford Chance, Crowell & Moring, Freshfields, Fried Frank) are building proprietary AI systems trained on internal firm data rather than relying on third-party legal AI vendors. The trend re-frames the AI buy-vs-build decision for the rest of the bar. None of the reporting addresses the privilege, confidentiality, or vendor-diligence questions the trend creates.
The lede. #
A summer 2026 wave of internal-data AI initiatives at Kirkland, Clifford Chance, Freshfields, Crowell & Moring, and Fried Frank reframes the BigLaw AI landscape around proprietary corpora; the governance questions that follow are unresolved.
What happened. #
In a June 7, 2026 Bloomberg Law report, multiple AmLaw 100 and global elite firms were named as building proprietary AI tools trained on internal-firm data. Kirkland & Ellis is the lead example with its $500 million Palantir-built platform (covered separately). Clifford Chance, Crowell & Moring, Freshfields, and Fried Frank are each described as developing institutional AI systems integrated with firm workflows and data. Fried Frank in particular is described as drawing on "a long history of well-structured data" for private funds clients. Polsinelli is identified as a firm pursuing third-party AI tools (Harvey, Legora) rather than building custom. The reporting frames the trend around competitive differentiation and alternative billing models.
Why it matters. #
The reporting omits what is, from a governance perspective, the central question: when a firm feeds client documents, matter files, deposition transcripts, and privileged communications into a proprietary AI system, how does the firm preserve the privilege, segregate access between matters, comply with conflict-of-interest duties, and meet its vendor-diligence obligations to the platform provider (Palantir for Kirkland; unspecified third parties for others)? The federal common-law analysis recently articulated in Heppner v. United States on consumer-tier AI privilege waiver applies, in modified form, to enterprise-tier internal AI as well — the question is whether the vendor's data-handling terms create an expectation of confidentiality that supports the Kovel-equivalent analysis. None of the public reporting addresses these questions. They will be addressed, eventually, in litigation.
What to watch. #
Three things. First, any firm that publishes its internal-data AI governance framework openly — that would be a first and would set a market reference. Second, the first piece of litigation in which an opposing party seeks discovery of internal-AI-system records as either privileged or non-privileged communications. Third, state-bar guidance specifically directed at internal-data AI systems; existing AI ethics opinions (California, Florida, NC, DC, ABA Op. 512) were drafted with off-the-shelf tools in mind and do not yet address proprietary corpora. For SMB and mid-market firms, the practical near-term work is the vendor-diligence framework around the off-the-shelf tools the firm actually uses. See AI vendor diligence: contract clauses to redline.
Primary sources. #
- Law firms capitalize on internal data to stand out in AI age, Bloomberg Law
- Kirkland $500M AI bid highlights access-to-justice issues, Bloomberg Law
- ABA Formal Opinion 512
- United States v. Kovel, 296 F.2d 918 (2d Cir. 1961) — functional equivalent doctrine
News brief. Not legal advice. Dan Hughes is not an attorney; IXSOR does not provide legal services.