AI adoption is no longer the hard part. Measuring it, governing it and scaling it are.
Legal AI has officially moved out of the novelty phase. Axiom’s preview of its 2026 In-House Legal AI Report, based on a March 2026 survey of 528 in-house legal leaders across six countries, finds that only 7 percent of teams have scaled AI across the organization, while 83 percent cannot measure whether their AI spending is working. Axiom’s own framing is blunt: “Legal AI is everywhere. Now comes the hard part.”
That hard part is no longer adoption. It is management. The report preview says nearly every team using AI expects to raise its AI budget next cycle, but very few have turned that investment into repeatable, measurable value. In other words, legal departments have entered the dangerous middle stage of AI maturity: tools are approved, budgets are moving, expectations are rising, but governance, workflow design and performance measurement are lagging behind.
For general counsel, the signal is clear. The AI conversation should shift from whether the legal department is “using AI” to whether it can prove what AI is doing. That means defining use cases, assigning ownership, documenting risk controls, measuring cycle-time improvement and creating an audit trail for vendor performance. A tool that cannot be measured is not a transformation strategy. It is a line item.
The legal departments that win with AI will not necessarily be the earliest adopters. They will be the ones that turn AI into managed infrastructure: governed, measurable and embedded in work.
Expect more pressure on legal ops, procurement and outside counsel management to prove ROI. AI budgets may rise, but tolerance for vague productivity claims will fall.
