New Research
AI Use Cases And Adoption In Corporate Legal Departments
New industry data shows generative AI use in corporate legal departments has nearly doubled in a year, and real use is showing up in contract review, e-discovery, and case strategy. New research on law firms shows something else: almost none of them can say what actually changed once the tools went live. Here’s where AI is actually being used, and the governance gap most firms still haven’t closed.
Eighty-seven percent of corporate legal departments now use generative AI (also called GenAI), nearly double the 44% reported a year earlier, according to data reported by PYMNTS on August 17, 2026. That is one of the fastest adoption curves we’ve seen in a professional services category. It is also, on its own, the least interesting number in this story.
The more important finding came a week and a half earlier, on August 6, when BARBRI published new research on how law firms are actually managing that adoption. Their answer: not well. Firms self-graded their own AI rollout as a C. Not one firm surveyed had built a formal AI competency framework for incoming associates. Firms can see who activated a tool. Almost none of them can see what changed in how that person actually works.
Key Takeaways
- 87% of corporate legal departments now use generative AI, up from 44% a year earlier (PYMNTS, Aug. 2026).
- The most common AI use cases in corporate legal are summarization (83%) and contract clause identification (63%), alongside e-discovery, document review, and contract drafting.
- New BARBRI research found law firms grade their own AI rollout a C, and no firm surveyed has built an AI competency framework for incoming associates.
- General counsel are shifting from evaluating AI tools for risk and return to governing systems capable of making independent decisions, often through AI-specific clauses added to vendor contracts.
- Tool activation and real behavioral change are two different things, and most firms are only tracking the first one.
- Every Simpatico employee completed AI training and signed our internal AI policy before we ever used an AI tool on a client system.
The Number Everyone’s Citing
The PYMNTS data is worth sitting with for a second. Legal departments went from 44% GenAI usage to 87% in twelve months. Adoption at that pace inside a profession built on precedent and caution is unusual, and it’s changing what general counsel actually spend their time on.
Reece Clark, a technology transactions attorney, framed the shift this way: in 2025, GCs were evaluating AI tools for risk and return. In 2026, they’re managing systems capable of making independent decisions without continuous human intervention. That’s a different job description, and a lot of legal departments are doing it without having updated the job description.
One place that shift is already visible: general counsel are increasingly using AI-specific addenda in vendor contracts as their primary governance tool, addressing liability, compliance violations, and errors made by autonomous systems. That’s a sign the contract is doing work that an internal policy should have already been doing.
Where AI Is Actually Being Used
Adoption numbers only tell half the story. The PYMNTS data also breaks down what legal departments are actually doing with generative AI once it’s live, and the pattern is consistent: the highest-adoption use cases sit closest to document work, not courtroom strategy.
Summarization
83% of legal departments use generative AI to summarize documents and communications, the single most common use case reported.
Contract Clause Identification
63% use AI to identify and flag specific contract clauses, cutting manual review time on standard agreements.
E-Discovery And Document Review
AI-assisted review is now a standard part of e-discovery workflows in corporate legal departments, not an experimental one.
Contract Drafting
Generative AI is also used to produce first drafts of standard contract language, with attorneys reviewing and finalizing.
Specialized legal AI platforms are seeing fast adoption too. Harvey, a legal-specific AI platform, is now used by roughly 70% of AmLaw 10 firms and nearly half of AmLaw 100 firms, serving over 700 customers across 58 countries. That’s the kind of adoption curve legal technology used to take a decade to reach, not a year. It’s part of the same broader shift in law firm technology strategy we work through with clients directly, not an isolated legal-department trend.
What Firms Aren’t Doing
The BARBRI research is the part of this story that should change how you’re planning the next quarter, not just how you’re reading the news. A legal innovation consultant quoted in the research put the underlying problem bluntly: “Either there are too many cooks in the kitchen, or nobody is managing it. It’s feast or famine.” Rollout is happening. Ownership of what happens after rollout usually isn’t.
Tool Activation Isn’t Fluency
Firms can report exactly who turned an AI tool on. Almost none can say what changed in how that person drafts, reviews, or advises as a result.
No Competency Framework
No firm surveyed has built a formal AI competency framework for incoming associates, meaning new hires are learning firm AI norms informally, if at all.
L&D And Innovation Are Disconnected
Collaboration between Learning & Development, Knowledge Management, and Innovation teams remains fragile, even where it’s improved since last year.
Vendor Contracts Are Filling The Gap
With internal policy lagging, general counsel are pushing governance work into vendor addenda covering liability and autonomous-system errors instead.
BARBRI Co-CEO Lucie Allen put the stakes on a longer clock: the firms that will lead by 2030 are the ones that recognize tool adoption is only part of the transformation required, and that real progress depends on aligning L&D, Knowledge Management, and Innovation to turn tool activation into actual fluency.
What We Did Before We Touched A Client System
We went through this ourselves before any Simpatico team member used an AI tool on client work. Every person on our team completed AI training and signed our internal AI policy first, not because a regulator required it, but because “trust the tool, skip the framework” is exactly how a firm ends up finding out the hard way what an AI agent did with client data.
We’ve also built AI into a live legal workflow ourselves. Simpatico built a Microsoft Copilot case management agent for a legal client, published as a case study by Pax8, Microsoft’s partner. The lesson from that build was the same one this research points to: the tool is the easy part. The policy, training, and human review process around it are what actually determine whether AI adoption reduces risk or just adds a new kind of it.
What To Do About It
Inventory What’s Actually Running
Know every AI tool active in your firm today, not just the ones IT formally approved. Shadow AI use is the norm, not the exception.
Write The Policy Before You Need It
A written AI policy should cover acceptable use, client data handling, confidentiality safeguards, and where human review is required.
Train Everyone Who Touches It
Not just associates. Paralegals, staff, and partners are all using these tools, and all need the same baseline training.
Assign Real Ownership
Governance needs a named owner across L&D, IT, and risk. Left to whoever activated the tool first, it doesn’t get owned at all.
Revisit It As The Tools Change
A policy written for chatbots doesn’t cover agents that take independent action. Plan to update it, not set it once.
Frequently Asked Questions
Do we need a written AI policy if we already require approved tools only?
What should a law firm AI policy actually cover?
Does adopting AI tools create new confidentiality risk?
How is an internal AI policy different from AI clauses in vendor contracts?
Where do we start if we have AI tools live and no policy?
Find Out Where Your Firm Stands
A conversation with our team on what’s actually running in your firm, what a real AI policy should cover, and where the gaps are before a client or regulator finds them.
- Your current AI tool inventory
- Policy and training gaps
- What to prioritize first
30 minutes · No pressure · No obligation
Share this article