Two attacks are absorbing the legal profession. They are nine months apart on the same calendar. They arrive from opposite directions and converge on the same end state. The first attack moves through the regulatory perimeter. The Arizona Supreme Court approved KPMG Law US on February 27, 2025, as the first Big Four-owned U.S. law firm. Four months later, the same court approved Eudia Counsel, an AI-augmented firm backed by $105 million from General Catalyst. By April 2025, Arizona had licensed 136 alternative business structure entities. California signed AB 931 in October to narrow the perimeter. Texas Opinion 706 validated the MSO concept in February. The attack is loud because it has to be. Through the click-through comes the second attack. On November 20, 2025, Norm Ai opened Norm Law in New York with a technology licensing model that requires no regulatory approval anywhere. On May 7, 2026, Legora launched what its investors call “foundational infrastructure” for legal work. There were no public hearings. No regulatory orders. No state bar approval. There was a press release, a terms of service, and a procurement decision. **One attack you saw. One attack you already signed.** **Two Attacks. One Calendar. One Destination.** Front door: the alternative business structure. A state bar admits a non-lawyer-owned entity to provide legal services. Arizona has been the test case since 2021, when the state eliminated Rule 5.4 and opened the regulatory perimeter. Utah ran a sandbox. Puerto Rico authorized 49% non-lawyer ownership in 2025. Texas approved the MSO concept in February 2025. California allowed flat-fee MSO arrangements in October while blocking contingency-fee ABS sharing through AB 931. The architecture is public, debated, and defended state by state. Open window: the technology licensing model. A non-lawyer-owned platform owns the technology, employs the engineers, and absorbs the data. The lawyer-owned firm pays a licensing fee for access. Capital sits above the firm without ever owning it. The architecture works in all fifty U.S. jurisdictions because no jurisdiction prohibits law firms from licensing technology. There is nothing to debate. There is nothing for a state bar to approve. **Front Door, Watched** The Loyalty Gap series previously documented the regulatory attack in detail. [See *The Loyalty Gap, Part 2: Rule 1.7 Disclosure*, The Technology Blind Spot (2026); *The MSO Barbarians at the Gate*, The Technology Blind Spot (2025); *The Private Equity Playbook*, The Technology Blind Spot (2025).] What is new in 2025 and 2026 is the speed and the volume. Holland & Knight published two separate practice guides on the MSO model in July and October 2025. Sidley Austin published a client advisory in November 2025 describing the trend as “a pivotal shift in the legal industry.” Reed Smith characterized MSO structures as having “moved from experimental concepts to repeatable, financeable platforms.” The Private Equity Legal Alliance, a consortium supporting PE deals in legal services, formed in late 2025. The vendors are publishing the playbook because the playbook is now standard. Catherine sees this attack. Her bar association has scheduled CLEs on the MSO model. Her partners have circulated the practice guides. Her clients have asked her about it. The press releases land in publications she reads. A managing partner who has not engaged with this attack has done so deliberately. **Open Window, Unwatched** The just-published Your Platform Has Three Owners. None of Them Are You. documented the technology licensing model in detail. [See Your Platform Has Three Owners. None of Them Are You., The Technology Blind Spot (2026).] Two facts from that analysis matter here. First, on November 20, 2025, Norm Ai opened Norm Law in New York under a technology licensing arrangement. Lawyers own Norm Law. Blackstone, Bain Capital, Coatue, and other institutional investors own Norm Ai, the platform. Norm Law pays Norm Ai for technology access. Per legal-tech analyst Mike Bommarito’s November 2025 analysis, this structure “can operate in any U.S. jurisdiction without special regulatory approval.” Per the Norm Ai press release, “Blackstone and Norm Ai are now also collaborating to shape and develop Norm Law legal services for Blackstone’s use.” Mike Schmidtberger, who chaired Sidley Austin’s executive committee from 2018 through 2025, joined Norm Law as chairman in February 2026. Second, on May 7, 2026, Legora announced its “agentic operating system” for legal work, complete with embedded “Legal Engineers” inside firms and end-to-end orchestration “from first-mile matter intake, through research, drafting, and review, to last-mile client delivery.” Accel, which led Legora’s $550 million Series D in March 2026, characterized the bet as “build the system that both lawyers and agents standardize upon, not a feature but foundational infrastructure.” This attack proceeds without engagement from the managing partner. The press releases land in legal-tech outlets, not bar association newsletters. The terms of service is a click-through that the firm’s IT lead approves during procurement. The state bar inquiry has not opened. The press cycle has not started. The architecture is in production at three firms before any state bar issues an opinion. **Asymmetry the Profession Has Not Named** The two attacks converge on the same end state: capital extracts rents from a market historically reserved for lawyer labor. The routes differ on six dimensions, and on most of them the open window is structurally worse than the front door. © THE TECHNOLOGY BLIND SPOT All Rights Reserved The Trojan Horse is silent because nothing about it requires a managing partner’s attention. There is no transaction document. There is no equity transfer. There is no state bar filing. There is a vendor procurement decision that, in many firms, the managing partner does not see. The associate signs the click-through. The IT lead approves the integration. The bookkeeper pays the invoice. The managing partner ratifies a decision she did not know she was making. Consider the managing partner who, in October 2024, declined to take a meeting with a private equity firm interested in an MSO conversation. She was firm. Rule 5.4 was the line. She would not consider it. In the same eighteen months, that managing partner signed her firm into Microsoft Copilot, two AI contract review platforms, and an enterprise legal-AI agent for matter intake. Three of the five most prominent venture-capital and private-equity investors in legal AI sit somewhere in the cap tables of those vendors. None of those investors knocked at the front door. Each of them came in through the click-through. Each privacy policy authorizes disclosure to government and to third parties in language similar to the Anthropic terms Judge Rakoff held preauthorize disclosure without compulsion. See United States v. Heppner, No. 1:25-cr-00503-JSR, ECF No. 27 (S.D.N.Y. Feb. 17, 2026) (Rakoff, J.). The capital she refused to admit through the front door is in the building. It came in through the open window. She did not refuse the capital. She refused the paperwork. **Where the Frame Breaks** The strongest counterargument to this framing is that not every Trojan Horse vendor climbs the stack. Some legal-AI vendors offer customer-controlled key management. Some publish transparency reports. Some operate on flat-fee licensing without revenue-tied structures. Some firms have negotiated MSA notification clauses requiring vendors to notify before complying with subpoenas. These architectures exist. That counterargument is correct as far as it goes. They are also the exception. The market direction, as documented in Accel’s investment thesis, FirstMark’s customer-behavior post, and Norm Law’s published structure, is the integrated platform, not the customer-controlled architecture. The vendors competing for the next round of funding are climbing the stack, not anchoring at the bottom. The firms that have negotiated favorable contracts are large enough to negotiate. Catherine runs a fourteen-attorney litigation practice. She did not negotiate. She accepted the click-through. The frame has a second limit. Some MSO arrangements have legitimately benefited under-capitalized firms, particularly plaintiff-side personal injury practices that lack capital for marketing infrastructure. Eudia Counsel’s stated mission of reducing in-house counsel reliance on outside firms represents a genuine value proposition for corporate legal departments. Not every regulatory-perimeter attack is predatory. Some firms have made informed, eyes-open decisions to engage with capital through the front door, the open window, or both. The frame is sharpest for procurement decisions still ahead. It is not judgment of arrangements made deliberately. **Three Questions by Thursday** Your Platform Has Three Owners prescribed three contract clauses to pull from a single vendor. This piece prescribes one binary question across three vendors. By Thursday, Catherine should pull contracts for the firm’s three highest-spend AI or technology vendors. For each contract, she should answer one question: does this arrangement, in operational substance, transfer to the vendor any of the rights an MSO would have transferred? First, control over operations. Test for embedded engineers who configure matter workflows, prescribed clause libraries the firm cannot override per matter, platform-level decision rules above attorney judgment, or vendor staff who participate in client communications. Each is the operational signature of an MSO arrangement renamed as a software license. Second, capture of margin. Test for pricing that scales with the firm’s matter volume, revenue, or specific case outcomes, directly or through usage-tier breakpoints calibrated to firm growth. Per-seat licensing falls outside this concern. Per-matter, per-document, or per-recovery pricing falls inside it. Watch for usage caps that recalibrate annually against the firm’s billings. Third, claim on data. Test for clauses reserving the vendor’s right to use firm inputs for model training, derivative outputs, training-corpus retention, or aggregated analytics shared with affiliates or investors. The clause may be silent on opt-out, or may permit opt-out only at the matter level rather than firm-wide. Either is the architectural signature of the data layer documented in Your Platform Has Three Owners. Fourth, share of revenue. Test for payment that California Rule 5.4 Comment [2] would describe as “a percentage or share of the lawyer’s or law firm’s overall revenues or tied to fees in particular cases.” Total spend that exceeds a meaningful share of firm revenue creates Rule 5.4(c) concern even at flat-fee structure: at sufficient scale, the vendor effectively directs the firm’s professional judgment by controlling operational infrastructure the firm cannot exit. If yes on any sub-test for any of the three vendors, the firm has signed a Trojan Horse arrangement and should know it. The arrangement may still be acceptable. It may still produce more value than it costs. But the firm should know what it has agreed to before any conflict, any government order, or any state bar inquiry arrives. The exposure is not theoretical. It is in the procurement file. **Knock and the Window** KPMG knocked. Eudia knocked. Texas opened the door for the MSO in February. California closed the ABS lane in October but kept the MSO lane open. The conversation about the regulatory perimeter is loud because the perimeter is the only part of the building the profession watches. While the profession watched the door, Norm Law walked through the open window. Legora has published the architectural template. Bommarito’s November 2025 analysis predicts imitators within eighteen months. The investors are betting on this becoming the standard. The strategists are charting it. Senior Big Law has stopped debating it. State bars are not yet looking. The press cycle has not started. The opinion that will name this structure does not yet exist. When it arrives, every firm now signing click-through terms of service will sit inside an arrangement no managing partner explicitly chose. Most managing partners did not refuse the capital. They refused the paperwork that would have made the capital visible. The architects published two maps. One they showed you. One they did not. Read both, or you cannot tell which surrender you have already signed.

Originally published on LinkedIn Newsletter: The Technology Blind Spot