AI Is Everywhere in Law, but Hiring Has Barely Moved

Most legal professionals say AI has not yet changed their hiring, and more report adding staff than cutting it, according to new research from Relativity and IDC.

AI in Legal: 2026 Research Data is a study from Relativity, a legal data intelligence company, with research and analysis by International Data Corporation (IDC), that examines how legal professionals are adopting and governing AI. Relativity released the study on 12 November 2026.

AI in Legal 2026 at a Glance

• Most respondents (57.9 per cent) said AI has had no impact on their hiring, while 24.2 per cent reported increased hiring and 9.2 per cent decreased hiring
• AI is used in legal work by 80.5 per cent of legal professionals and more than 91.0 per cent of chief legal officers
• More than 99.0 per cent of respondents either use AI now or plan to within 12 months
• Respondents favoured accuracy over speed (92.3 per cent to 7.7 per cent) and higher trust over lower cost (79.3 per cent to 20.7 per cent)
• The billable hour remains the most common primary billing method (32.5 per cent)
• Only 23.0 per cent of organisations have fully adopted an AI governance framework

What Is AI in Legal: 2026 Research Data?

The study combines IDC primary research, a survey of state court judges conducted by Relativity and the National Association of Women Judges (NAWJ), and findings from the upcoming eighth annual edition of The General Counsel Report, produced by Relativity and FTI Consulting. Together, these perspectives offer a broad view of AI use across the legal sector, along with the growing importance of trust, governance and accountability.

The research was conducted from April to October 2026 and incorporates feedback from 700 legal professionals across 20 countries, including an IDC study of 401 respondents. Participants included lawyers, paralegals, legal technologists and legal IT professionals.

The countries represented are Argentina, Australia, Brazil, Canada, China (Hong Kong), Colombia, France, Germany, India, Ireland, Japan, the Netherlands, New Zealand, Singapore, Spain, Sweden, Switzerland, the United Arab Emirates, the United Kingdom and the United States.

‘In this 2026 follow-up to our 2024 study, the data show that AI use in the legal profession is now a fait accompli. The question is no longer "Are legal teams using AI?" but "How and why are lawyers and their colleagues using it?"’ said David Horrigan, discovery counsel and legal education director at Relativity.

‘Some of the data, such as more than 91.0 per cent of chief legal officers and more than 80.0 per cent of all legal professionals using AI, was not surprising at all. What did surprise us was that, for most respondents, AI has had no effect on hiring at this point, and more reported adding staff than cutting it.’

Has AI Changed Legal Hiring and Billing?

The data suggest AI is not yet driving widespread staffing reductions. According to IDC, 57.9 per cent of respondents said AI has had no impact on their hiring, and more reported increasing hiring (24.2 per cent) than decreasing it (9.2 per cent).

‘Many have predicted that AI will end the billable hour on the theory that faster work will lead clients to demand task-based billing,’ said Horrigan. ‘The research indicates that has not happened broadly, at least not yet.’

The billable hour remains the most common primary billing method among respondents (32.5 per cent), ahead of task-based billing (22.3 per cent), hybrid billing (20.3 per cent) and contingency fees (18.0 per cent).

The research also points to cost savings, with 82.8 per cent of respondents saying AI has saved their organisation money.

How Widely Is AI Used in Legal Work?

The research indicates 80.5 per cent of legal professionals and more than 91.0 per cent of chief legal officers currently use AI in their legal work, and more than 99.0 per cent of respondents either use it now or plan to within 12 months. Usage is also growing, with 93.0 per cent of respondents saying they have increased their AI use by an average of 65.0 per cent, while just 0.9 per cent have decreased it.

‘When adoption ceilings sit above 90.0 per cent, using AI stops being a differentiator and starts being a floor,’ said Ryan O’Leary, research director for privacy and legal technology at IDC. ‘The advantage moves to how well it is used.’

How Do Chief Legal Officers and Judges Differ on AI?

According to data from The General Counsel Report, 86.8 per cent of chief legal officers are very or somewhat optimistic about AI.

Judges, by contrast, are adopting AI more slowly. Data from the Relativity and NAWJ survey indicate 16.6 per cent of respondents never use AI and another 33.3 per cent use it only rarely. About one in five federal and state judges (20.4 per cent and 20.0 per cent respectively) prohibit court staff from using AI, while 7.4 per cent of federal judges and 14.2 per cent of state judges permit and encourage it.

Judges’ outlook is also more divided. In the Relativity and NAWJ survey, 47.1 per cent of state court judges said they are very or somewhat optimistic, and 36.0 per cent said they are very or somewhat concerned. Among federal judges surveyed by The Sedona Conference, 43.6 per cent were optimistic and 41.7 per cent were concerned.

Why Do Accuracy and Trust Outrank Speed and Cost?

In the IDC research, trust consistently outweighed speed and cost. When forced to choose, respondents favoured accuracy over speed (92.3 per cent to 7.7 per cent) and higher trust over lower cost (79.3 per cent to 20.7 per cent). Respondents also preferred keeping a human in the loop (64.1 per cent) over complete automation (35.9 per cent).

As AI agents enter legal workflows, the study indicates that oversight will determine whether trust keeps pace with adoption. For agentic AI, respondents said human approval (67.0 per cent) and mandatory source citations (60.0 per cent) are the most commonly required safeguards.

The top three qualities legal professionals look for in AI vendors are:

• Security posture (59.9 per cent)
• Governance capabilities (59.4 per cent)
• Clear AI principles (51.9 per cent)

‘Buyers aren’t shopping for features first,’ said O’Leary. ‘They’re checking whether you’re safe to let near their data.’

Where Does AI Governance Stand?

Inside organisations, governance work is still under way. About 91.0 per cent have an AI governance framework in place or in progress, yet only 23.0 per cent have fully adopted one, and 9.0 per cent have no plans at all, according to IDC’s research.

Concerns about accuracy and data security remain the gatekeepers to deeper adoption. Hallucinations, which were not a leading concern in the 2024 study, are now respondents’ top worry (52.0 per cent), followed by data loss (46.0 per cent) and data bias (41.0 per cent).

Yet hallucinations are also the risk legal professionals feel best equipped to handle. Of those surveyed, 44.7 per cent say they are almost or fully prepared for them, and only 15.5 per cent say they are not prepared at all.

Why It Matters

For law firm leaders and legal IT teams, the findings suggest the question has moved from whether to adopt AI to how well it is governed and used. With security posture and governance capabilities ranked above other qualities in vendor selection, and fewer than one in four organisations having fully adopted a governance framework, governance maturity may become a sharper point of difference than access to the tools themselves. The persistence of the billable hour also suggests that, so far, efficiency gains have not translated into broad changes to how legal work is priced.

The full study, AI in Legal: 2026 Research Data, is available for download.