AI is advancing, but the gap between hype and adoption is still stark.
Doug Austin’s 2026 State of the Industry Report landed with some eye-opening numbers: 60.7% of the 559 of the respondents expect LLMs and GenAI to be transformative by the end of the year. Although this is nothing new, the gap between hype and actual usage tells a more interesting story about where ediscovery really stands right now, and what you can do to stay prepared. Make sure you subscribe to eDiscovery Today to get the full report.
The AI Adoption Gap
When asked if GenAI is already transformative, 23.3.% of survey respondents said yes, while another 37.4% expect it to reach that status this year. Add those together and you’ve got more than three in five practitioners who believe we’re living through a fundamental shift.
But only 17.7% are actually using LLMs/GenAI in all or most of their cases.
Doug calls this out in the report: “Riddle me this, Batman! Why did 23.3% of respondents say LLMs/GenAI is already transformative, when only 17.7% of respondents said they use LLMs/GenAI technologies in all or most of their cases? Hype maybe?”
This gap tells us we’re still early on the path to AI adoption — not at the finish line. The question for legal teams is how to move from just dabbling in it to doing it right.
How People Are Actually Using AI
When asked which use cases they’re applying LLMs/GenAI in actual projects and cases, the respondents selected these use cases:
- Document Review (66.2%)
- Document Summarization (65.8%)
- Case Strategy/ECA (52.4%)
- Privilege/PII Identification (49.9%)
- Timelines of Key Events (47.0%)
- Translation (44.0%)
- Deposition Prep/Summaries (40.8%)
- Other (5.0%)
- None (2.9%)
More than 30% of respondents selected all seven use cases listed, showing that AI adoption has moved well beyond early testing into everyday workflows.
Kelly Twigger, Principal at ESI Attorneys and CEO of Minerva26, sees AI’s potential beyond the obvious use cases.
“Everyone talks about review acceleration; fewer talk about preventing the dispute that creates the review,” she said. “The move is to use LLMs to draft and stress-test the decisions — scope, preservation, collection posture, sampling, and privilege approach — then make humans own the reasoning and tie it to the record.”
The range of use cases shows AI is moving beyond just speeding up review. Legal teams are applying it across the entire discovery lifecycle, from early case assessment through deposition prep.
What Are the Roadblocks to Adoption?
The roadblocks aren’t really about the technology itself anymore. When asked in the 2026 State of the Industry Report about the single biggest roadblock to LLM and GenAI adoption, respondents pointed to:
- Cultural resistance and skepticism
- Not understanding how the technology actually works
- Defensibility concerns and lack of case law
- Worries about hallucinations and accuracy
Hon. Xavier Rodriguez of the Western District of Texas, who Nextpoint recently partnered with on a webinar about detecting deepfake evidence in litigation, put his response bluntly: We should expect, “regrettably, more discovery disputes as to transparency of the process and verification of outputs.”
Kelly nailed a core issue:
“The failure mode is speed without receipts,” she said. “The move is grounding, logging where appropriate, validation, and human accountability so strategy scales across matters without collapsing credibility.” Speed without defensibility doesn’t help anyone.
The Problems We Still Haven’t Solved
While everyone’s talking about AI, two fundamental ediscovery challenges are still tripping people up:
Mobile Device Discovery is basically stuck. When asked how often they discover mobile devices, only 38.0% of respondents said in all or most cases, while 20.6% said in very few or none of their cases. The problems cited were privacy concerns, BYOD headaches, and the technical difficulty of doing it defensibly.
Collaboration Apps, such as Slack and MS Teams, have received more attention. 42.5% of respondents said they discover collaboration app data in all or most cases, while only 15.2% rarely or never do. Government entities show different patterns, with 36.7% rarely or never collecting this data. This gap likely reflects both lower necessity and the budget constraints common in government agencies.
Mary Mack, CEO of EDRM, summarized the challenge in her response: “The mix of personal with business, and the cost in dollars and time to disentangle it.”
The Hyperlinked File Question
Should hyperlinked files be treated as “modern attachments” and produced along with emails? When asked this question, the responses show the industry wrestling with consensus:
- 40.5% said yes, but only if the parties negotiate it
- 41.5% said produce them if responsive or always produce them
- 18.0% said produce them with relevant emails by default
Kelly Twigger says, “The endgame is the record you need for summary judgement and, ultimately, jury instructions.” If hyperlinked content matters to your case, you need to address it in your ESI protocol upfront, rather than hoping it just shows up in a production.
What We’re Still Not Talking About
When asked in the 2026 State of the Industry Report what ediscovery challenge isn’t being talked about enough, for the sixth year running, “Lack of eDiscovery Competence within Legal” topped the list at 22.2% of responses. But this year had a new runner-up: “Evidence Authentication Issues Due to Rise of Deepfakes” at 15.0%.
Among government entity respondents specifically, nearly half (48.5%) identified competence as the biggest unaddressed issue. Disparities in access to ediscovery resources have diminished as ESI proliferates all types of litigation, but a gap still remains between smaller matters and big, high-stakes cases.
Judge Rodriguez noted this discrepancy in his response. “Too often we contemplate ediscovery challenges in the ‘big’ case context,” he said. “Most state and federal court cases don’t have those monetary stakes at play. Yet, cell phone, chat, and collaborative app data is still needed to prove or defend in these cases.”
This competence gap can impact conversations with clients, and ultimately, a firm’s bottom line. Brett Burney, Nextpoint’s eLaw Evangelist, highlighted another under-discussed ediscovery challenge:
“Lawyers failing to help clients comprehend the expense involved with conducting a proper ediscovery review and production.” He continued, “Maybe lawyers don’t fully understand it all themselves so they’re reluctant to try and explain it to clients. They write-off the expenses when it’s something they can ethically charge for and should.”
The rise of deepfakes as the second-most cited challenge adds a new dimension to this competence issue. Lawyers now need to understand not just how to collect evidence, but how to authenticate it in an era of AI-generated content. If you want to dive deeper into this emerging challenge, Nextpoint and EDRM recently hosted a webinar with Judge Rodriguez, forensic expert Dr. Felix Hernandez, Brett Burney, and Doug Austin on how to detect and challenge deepfake evidence in the courtroom. Watch it here for practical insights on spotting and addressing deepfake evidence in litigation.
What Legal Teams Should Actually Do
Based on the 2026 State of the Industry report, here’s what matters most for teams that want to stay ahead:
1. Move from Pilots to Policies
2026 is the year to move from speculation and excitement to action with AI. Document your workflows, set up validation procedures, and create governance frameworks that can stand up to scrutiny in discovery disputes.
2. Fix the Fundamentals
Don’t let shiny AI objects distract you from the basics. Get your mobile device collection strategy figured out. Standardize how you handle collaboration apps. Decide your position on hyperlinked files before you’re arguing about it in court.
3. Invest in Competence
Whether it’s training, consultants, or better technology partnerships, make sure your team understands the technology tools available and the discovery principles underneath them.
4. Build for Defensibility
Kelly states, “Without provenance and validation, AI becomes a credibility risk, particularly around privilege and confidentiality.” Every AI-assisted decision needs documentation that explains the reasoning behind it.
5. Talk About Money
Brett’s point about lawyers struggling to explain ediscovery costs matters even more as AI changes the game. Clients need to understand what they’re paying for and why it’s worth it.
The Reality Check
Doug Austin’s report, now in its sixth year, shows an industry at a turning point. GenAI is here, people are using it, and the question is whether you’re using it well.
Doug wraps up the report, exclaiming, “eDiscovery continues to be more interesting, challenging and exciting than ever!” For legal teams willing to do the work of thoughtful, defensible implementation, 2026 could genuinely be transformative.
Watch Brett Burney and Doug Austin discuss the full State of the Industry Report in the latest Key Discovery Points video below:
Ready to Move Beyond AI Hype?
Get your copy of “Artificial Intelligence for the Rest of Us,” Nextpoint’s practical playbook for using AI effectively in legal work. This book offers an honest exploration of AI in the legal field — a foundational resource for teams navigating this rapidly evolving technology.
