ILTA 2026 Technology Survey Results Released: Legal Technology Trends

ILTA's 94% GenAI "adoption" number is softer than it looks — ask what your vendor counts as using it

The International Legal Technology Association released its 2026 Technology Survey this weekend, per eDiscovery Today's Doug Austin — a 489-page report (up from 445 pages last year) built from responses by 508 firms representing 139,661 attorneys and roughly 275,003 total users. The headline figure: 94% of firms are "using/exploring" generative AI, up 14 points on last year, ranging from 85% of firms under 50 lawyers to 100% of firms with 700 or more lawyers. Underneath that, 79% use GenAI for first drafts, only 13% report using agentic AI, 78% have a formal GenAI policy, 60% expect the profession to look "significantly different" in ten years, and accuracy concern fell to 63% from 82% the year before. The full report costs $500 for ILTA members and $800 for non-members unless your firm already participated or sponsors ILTA.

Does 94% mean firms are actually running GenAI on real matters?

No — the survey question merges "using" with "exploring," so pilot projects and one-off ChatGPT logins count the same as production deployment.

That distinction matters more in litigation than in most practice areas, because "exploring" a tool on a sandbox document is a very different risk posture than running it against privileged collections. The 79% figure for GenAI drafting is a cleaner signal of real usage than the 94% headline, and it's worth asking any vendor citing this survey which number they mean.

Why did the accuracy worry drop from 82% to 63% in one year?

The survey doesn't say — it could reflect better tools, more experience managing hallucination risk, or simply desensitization after a year of use.

A 19-point swing in "biggest concern" answers is large enough to deserve a mechanism, and ILTA's summary doesn't offer one. Before treating declining accuracy concern as evidence that GenAI got more reliable, ask whether respondents changed how they test outputs, or just got used to living with the risk.

Does firm size still gate what AI a firm can responsibly run?

Yes — adoption ranges from 85% at firms under 50 lawyers to 100% at firms of 700+, a real capability gap, not noise.

Only 13% of respondents overall report agentic AI in use, and that figure almost certainly skews toward the large-firm end of the range given the infrastructure and governance agentic workflows demand. If you're at a smaller firm evaluating an eDiscovery platform's AI claims, ask for adoption data broken out by firm size, not the blended 94%.

What should you ask your own eDiscovery vendor after reading this?

Ask how they define "using" GenAI, whether privileged data trains any model, and what their accuracy testing looks like — not just their adoption stat.

With 78% of firms now reporting a formal GenAI policy, procurement conversations are shifting from "do you have AI" to "does your AI meet our policy." A defensible platform should be able to answer chain-of-custody and data-isolation questions directly, rather than pointing to industry-wide adoption numbers as reassurance.

Frequently asked questions

Is the ILTA 2026 report free?

Only for member firms that participated, 2026 ILTA Corporate members, or paid ILTA365/Silver/Gold/Platinum/Titanium/ILTAMAX sponsors. Everyone else pays $500 (ILTA members) or $800 (non-members).

How much bigger is this year's survey than last year's?

The full report grew from 445 to 489 pages, though the Executive Summary shrank slightly from 13 pages to 11.

What section covers GenAI specifically?

The Emerging Technologies section, starting on page 47 of the PDF, with 35 tracked questions covering GenAI and agentic AI use.

Source: eDiscovery Today ("ILTA 2026 Technology Survey Results Released: Legal Technology Trends," Doug Austin, September 14, 2026).