Legal tech · Industry brief
Top three stories shaping Legal tech today, written for someone who already works in the industry: regulation, M&A, new entrants, notable filings, and any precedent worth pulling. Cite the trade publication (e.g. trade press, government source, court docket) directly so I can follow up.
← Zur aktuellen AusgabeLegal AI is mostly prompt layers, PE floods in, hallucinations spike
Legal AI market reality check
Most legal AI vendors are just wrapping general models.
HAQQ tested the market instead of reading marketing slides and found 65% of legal AI vendors are prompt layers over somebody else's foundational model, with only 3% trained on legal data [Source: HAQQ]. Large firms stack multiple AI systems rather than consolidate on one—80% run several tools or build their own router. More troubling: hallucination is endemic across every tested tool, with measured error rates between 17% and 33% for grounded legal research, and AI-fabricated court filings spiked from roughly 200 cases in September 2025 to 2,044 by September 2026.
FTC enforcement is coming—the agency already fined DoNotPay $193,000 for unsubstantiated AI performance claims.
Private equity raids law firm M&A
Law firms are the last professional services frontier without PE.
Following the Management Services Organization playbook—which separates operational infrastructure from legal practice—PE firms are racing into law firm consolidation [Source: Legal 500]. Recent anchor deals include New Mountain's stake in EisnerAmper and Grant Thornton, Apax's CohnReznick investment, and Blackstone's play in Citrin Cooperman. On the law firm side, precedent deals span Hogan Lovells/Cadwalader, Perkins Coie/Ashurst, and McDermott Will & Emery/Schulte Roth & Zabel.
Mid-Am Law 200 firms facing succession crises are the likeliest first movers.
OpenAI locks legal AI distribution
General model providers now control specialized legal distribution.
OpenAI's Astra for Law, launched September 17, ships 26 partner plugins that integrate iManage, Thomson Reuters HighQ, LegalZoom, and DeepJudge into a single research workbench, addressing the earlier Leah launch with platform scale rather than point solutions. The move signals that moat has shifted from models to integrations and data—OpenAI's CourtListener partnership indexes 230 million URLs covering 99.9% of U.S. precedential decisions. Specialized legal vendors now compete for plugin slots rather than standalone market share.
Watch whether mid-market firms abandon dedicated research tools for plugin stacks.
The Legal AI Landscape in 2026: What It Is Actually Made Of - HAQQ3 hours ago ... Legaltech Hub counts more than a thousand generative AI solutions in its directory. Venture funds publish tidy category grids. Review sites rank four hundred ...haqq.ai