Want more content like this? Sign up for our monthly newsletter.
The legal industry spent the last few years debating whether to use AI. Now that 92% of legal professionals report using AI for legal work, it turns out that was the easy part.
Getting there wasn’t a straight line, with adoption actually dipping from 74% to 69% between 2024 and 2025. What broke the standstill was experience: the professionals who kept going built enough firsthand evidence to pull the skeptics in. By 2026, 92% say the benefits of AI outweigh the risks.
So we asked three legal leaders what this year’s State of AI in Legal findings look like from where they’re standing. Jasmine Singh, Ironclad’s General Counsel, sat down with Mary O’Carroll, CEO of Legal Eng Consulting Group; John LaBarre, General Counsel at Harvey; and Irene Liu, Executive Director of Stanford Law’s AI Initiative and founder of Hypergrowth GC.
Here’s what they said.
How legal teams are actually using AI
Hearing that 92% of legal teams use AI is an impressive stat, but the term ‘using AI’ can mean a lot of things. There’s a spectrum that runs from occasional ChatGPT queries to orchestrating fleets of specialized agents.
Liu coaches a number of general counsels and sees the variance up close. “We can guarantee that 92% are all using chatbots for sure,” she said. And for most teams, that’s roughly where it stops. “A lot of them are not necessarily utilizing all the different types of basic functionalities like brainstorming with voice mode or using deep research functions to create more in-depth analysis.”
LaBarre sees the same gap, and frames it in terms of what most teams have done versus what’s actually possible. “The primary modality for using these tools has been: I know how I do this thing, I have these steps, and AI can make parts of these steps more efficient. It’s giving me bionic superpowers.” But he’s pushing for the industry to think bigger. “What if we start to just break the mold and say, how can I reimagine how we do this task using AI? I see 2026 as the year where we go from just improving the way we do our day-to-day to reimagining how we can deliver legal services.”
O’Carroll is direct about where most teams—including experienced ones—currently stand. “When it comes down to it, everyone I talk to is behind. We’re all behind. Everyone is in the pilot stage.” On what to do about it: “People tell me you shouldn’t use the word ‘take risks’ with lawyers. No, you should. You should take more risks and just try things.”
AI isn’t cutting back on legal work, and it never will
It’d be great if more capable tools meant less work. What actually happens, though, is that organizations don’t bank efficiency gains. Instead, the bar moves up, and the team has to rise to meet new expectations. O’Carroll puts it plainly:
The idea that workloads will go down because of technology has always been a myth.
MAry O’CarrollCEO, Legal engineering consulting group
AI might not be shortening the work week, but it is changing the work itself. The report found that 89% of legal AI users say they’re spending more time on complex, strategic work. Liu sees AI’s role in that shift less as a burden and more as an upgrade. “AI has become so powerful now that it’s far better than a junior intern. You could really use it as your chief of staff in many ways.”
O’Carroll described the legal ops role moving toward “sitting at the intersection of law, process reengineering, and translating all of that into technology.”
The skills that matter most in that environment, Singh added, aren’t the ones AI is best at replacing. “Good judgment, curiosity, and relational awareness—those are the things that make people in our profession even better at what they do in this moment where technology can take out the repeatable work.”
Legal’s job outlook is getting more optimistic
65% of legal professionals now believe AI will create more job opportunities, up from 43% in 2024.
What could be behind this rise are new types of jobs, and not necessarily a higher volume of them. Legal engineers, legal solutions specialists, and change management leads are all roles that require both legal fluency and systems thinking, and that former practicing attorneys are increasingly filling.
LaBarre points to his own team as evidence. At Harvey, LaBarre’s legal engineering team is larger than his GC team. “Those are all people who were practicing attorneys and are now helping with the process of change management. You can’t have a generative AI legal tool and not also offer change management as part of the offering.”
Liu sees the same shift from her vantage point at Stanford. “GenAI has really bolstered legal tech in many ways. There is a lot of excitement about what the technology is bringing to the profession. What’s cool is that you could still keep the job and try to do more strategic work.”
For Singh, the optimism stat put words to something she’d been feeling across the legal community.
So many of us years ago would lament: it’s hard to be a cost center. How do I position myself as a strategic thought partner? It felt like an uphill battle. And now we are the coveted.
Jasmine singhGeneral counsel, ironclad
98% of legal professionals say people outside their function have asked about their AI usage, whether that’s finance, sales, or the C-suite. Legal, whether it planned to or not, has become the organization’s go-to on the subject.
Everyone wants accountability, but most haven’t written it down
96% of legal professionals say they’d use AI more extensively if accountability for errors was more clearly defined. But only 49% have a formal error policy in place. Everyone else living in that gap might have trouble bridging the gap between conversations and formal guidelines. It’s all still new territory.
LaBarre traces the hesitation to something structural. “For the last fifty years, we’ve all interfaced with technology in the same basic way. You go to your calculator, you hit equal, and you know with absolute certainty you’re getting the correct answer. Now we have technology where that’s not how it works. You have to bring your own lens to it.”
The accountability gap shapes legal’s credibility, for better or worse. Organizations with clearly defined AI error policies are 36% more likely to rate legal as ahead of other functions on AI. Governance, more than usage, may be what separates the teams that are actually ahead from the ones that just seem that way.
When an error does happen, Liu’s advice is to treat it the way you would any other failure.
You can’t outsource the blame, and you have to be competent for your own duties. If there is an error, do a postmortem. Figure out what was the failure point, how can we fix those issues, and how can we add guardrails.
irene liufounder & CEO, hypergrowthGC
O’Carroll’s frame for day-to-day accountability is simpler. “Think about this as a young new person on your team. You have to look over their work, and you have to check it. They’re there to help you, but you are still accountable.”
What’s next
Legal has largely cleared the first hurdle: getting people to actually use AI. Ironclad’s State of AI in Legal report covers what’s next for a function that’s adopted faster than it’s formalized, and what closing that gap actually looks like. Download it today.
Ironclad is not a law firm, and this post does not constitute or contain legal advice. To evaluate the accuracy, sufficiency, or reliability of the ideas and guidance reflected here, or the applicability of these materials to your business, you should consult with a licensed attorney.
