Your Briefing: What's Happening in Legal AI Right Now

Google is now in BigLaw. The NYC Bar has published its first AI framework for transactional lawyers. And in-house teams are openly telling firms they plan to send less work their way. Here's what mattered last month.

BY THE NUMBERS

4  Big Tech companies now with dedicated legal AI products: Anthropic, Microsoft, OpenAI, and Google

94%  of in-house legal ops teams expect AI to enable them to bring more work in-house, per Wolters Kluwer legal ops survey

83%  of in-house legal AI teams cannot formally measure or track ROI, per Axiom's 2026 In-House Legal AI Report

5  factors the NYC Bar proposes lawyers weigh before using AI on any transactional document 

Sources: Google Cloud press release (Aug 25, 2026)  ·  Wolters Kluwer legal ops survey (2026)  ·  RollOnFriday In-House Lawyer Survey 2026  ·  NYC Bar Emerging Companies and Venture Capital Committee (Aug 2026)

TOP STORIES

Tools Worth Watching

Google is now in the legal AI market, and its launch partners are some of the most sophisticated firms in the world

On August 25, Google Cloud launched Gemini Enterprise for Legal, its first industry-specific packaging of the Gemini Enterprise platform. The product ships with purpose-built legal skills, pre-built agents, and connectors into the document management, e-discovery, and legal research systems firms already run. Four firms are named as launch customers: Cleary Gottlieb, Freshfields, Weil, and Williams and Connolly. That list is not accidental. These are large, sophisticated buyers with existing legal AI commitments, and their agreement to be named publicly is a signal that Google worked to earn.

The product covers contract review, due diligence, regulatory monitoring, and privacy requests, with a permissions model that automatically inherits a firm's existing ethical walls through its DMS connectors. Data, playbooks, client files, and negotiated positions stay within the organization's private cloud perimeter, and base models are not trained on customer data. Weil's Chief AI and Innovation Officer, Andrew Simon, described the firm's interest as extending across the broader Google ecosystem, where the firm can deploy a range of tools and build new applications, while still ultimately producing a lawyer's final work product in Word. That last point matters: the document layer is still where transactional deals land, and the platforms sitting above it are competing for the workflow that feeds it.

Platforms like Gemini Enterprise for Legal surface a firm's institutional knowledge at the workflow level: matter history, prior agreements, context from the DMS. But for transactional lawyers, institutional knowledge is most valuable at the moment of need, in the document, as a clause is being drafted or negotiated. That requires a different kind of AI, one that connects to a firm's complete deal history and delivers it at the point where the work actually happens.

→ Google Cloud: Gemini Enterprise for Legal launch announcement

→ Artificial Lawyer: Google, Weil and Gemini Enterprise for Legal

Governance

The NYC Bar publishes the first AI framework built specifically for transactional lawyers

The Emerging Companies and Venture Capital Committee of the New York City Bar Association published an AI policy paper on August 12, 2026, that is worth reading carefully if you’re a transactional lawyer. Unlike most AI guidance from bar associations over the past two years, which has focused on disclosure obligations and hallucination risk in court filings, this paper is directed at transactional practice. It concludes that AI tools may assist in legal work but cannot substitute for professional legal judgment, and specifically maps those obligations onto VC and emerging company documents.

At the center of the paper is a five-factor framework for deciding whether AI should draft or review a given document: how standardized it is, how complex the substance, how bespoke versus interchangeable it is across matters, how much experience the firm has with that document type, and how much negotiation is expected. The paper applies that framework to specific document types, with basic formation documents and NDAs rated highly for AI suitability, and M&A agreements rated low. The call for a national framework embedded in the ABA Model Rules is unlikely to move quickly, but the five-factor test is immediately usable as an internal evaluation tool, and the mapping onto confidentiality, supervision, and billing obligations under the existing Model Rules is practically useful now.

→ NYC Bar: Policy Paper on AI Tools in Emerging Companies and Venture Capital

→ Clio: Beyond the NYC Bar's AI Framework: What Comes Next for Transactional Lawyers

Business of Law

In-house legal teams are pulling work back from firms, and AI is what's making it possible

A convergence of data points from mid-2026 tells a consistent story: in-house legal teams are using AI to pull work back in-house, and they are telling their law firms about it. A majority of respondents in the RollOnFriday In-House Lawyer Survey 2026 predicted they will send fewer instructions to law firms over the next two years as a direct result of increased AI use. A GC in banking put it plainly: external counsel will be needed for fewer and fewer things because AI allows self-service at a fraction of the time and cost. A Wolters Kluwer legal ops survey found 94% of respondents expect generative AI will enable them to bring more work in-house, with 87% expecting wider AI adoption to reduce their law firm spend.

For transactional practices, the question this raises is specific: which categories of work are most at risk of moving in-house, and which are genuinely protected by the complexity and judgment they require? The data suggest the answer tracks closely with the NYC Bar's five-factor framework covered above. Standardized, repeatable documents, NDAs, board minutes, and routine formation work are the most exposed. Complex, bespoke, heavily negotiated matters, M&A agreements, financing documents, and shareholder agreements remain firmly in the protected category. The firms best positioned are those that have a clear point of view on where they add irreplaceable value and actively communicate it to clients rather than waiting for the conversation to arrive uninvited.

The subtler implication is for pricing. In-house teams are not asking firms to work for less. What they increasingly expect is pricing that reflects outcomes and value in an AI-enabled world. Fixed fees for commodity work and premium rates for the bespoke judgment that genuinely cannot be replicated in-house are the model that makes sense in this environment. Firms that are still billing hourly for work that in-house teams can now automate are the ones most likely to find those conversations becoming uncomfortable.

→ RollOnFriday: In-house lawyers warn they're swapping firms for AI

→ Wolters Kluwer: Insourcing vs. outsourcing: A strategic roadmap for legal ops