AI in law: The shift from fake quotes to real workflows

Law firms are finally moving past the embarrassment of AI-generated fake case citations. Tools like Casetext’s CoCounsel and Harvey AI are now being used for legal research, document analysis, and contract review—tasks that actually save billable hours. The shift isn’t just about avoiding PR disasters; it’s about firms quietly integrating AI into workflows where it can reduce errors and cut costs.

But don’t mistake this for a revolution. The adoption is still piecemeal, with most firms testing AI in low-risk areas like due diligence or e-discovery. A 2023 Thomson Reuters survey found only 15% of law firms had adopted generative AI, though 60% were exploring it. The real story isn’t how many firms are using AI—it’s how few are using it for anything beyond basic automation.

The hype cycle hasn’t disappeared; it’s just been repackaged. Vendors now sell AI as a compliance tool, a contract reviewer, or a predictive analytics engine. The pitch is less about replacing lawyers and more about making them slightly less inefficient. For now, that’s enough to justify the investment—but only if the tools deliver on the demo promises.

The competitive advantage isn’t in having AI; it’s in using it well. Firms that integrate AI into core workflows—like Latham & Watkins’ AI task force—are seeing measurable gains in speed and accuracy. Meanwhile, smaller firms risk falling behind if they can’t afford the tools or the training to use them effectively. The market is splitting into those who treat AI as a strategic asset and those who see it as a checkbox.

Developer activity tells a different story. GitHub repositories for legal AI tools are growing, but most are still experimental. The OpenLLM leaderboard shows legal-specific models lagging behind general-purpose ones, suggesting the industry is still in the early stages of customization. That’s a problem for firms expecting off-the-shelf solutions to solve complex legal problems.

The reality gap is widening. AI can draft a contract clause in seconds, but it can’t yet navigate the nuances of a high-stakes negotiation. Firms that over-rely on AI for substantive work risk compliance failures or client pushback. The smart money is on tools that augment lawyers, not replace them—at least until the next wave of hype arrives.

The real signal here is that AI adoption in law is no longer about flashy demos—it’s about incremental improvements in workflows. Firms that focus on measurable outcomes, like reduced review times or fewer errors, will outpace those chasing the latest AI buzzword. The tools are getting better, but the bottleneck remains human oversight.