The Legal AI Transactional Associate Workflow Audit 2026: What Mid-Level Associates at AmLaw 200 Firms Are Actually Using AI For, How Much Time It's Saving, and Where the Supervision Structure Is Failing to Keep Up

Research briefing — a review of public sources. The Legal Stack did not conduct a survey, benchmark or interviews for this piece.

Summary

We found no public study that measures, task by task, what mid-level transactional associates at AmLaw 200 firms use AI for or how much time it saves them. Law firms don't publish that data, and the available surveys cover lawyers in general rather than transactional associates. This briefing does not fill that gap with estimates. It does three things instead:

sets out what published surveys say about large-firm AI adoption and expected time savings;
uses documented court episodes to show where AI review chains have broken, at large firms included;
explains what the ethics rules require of supervisors.

The analysis of transactional workflows is labelled as such.

What the adoption data shows

Large firms lead adoption. The ABA's 2024 Artificial Intelligence TechReport (survey fielded October to December 2024) found that 30.2% of attorneys said their offices use AI-based technology tools. The figure rose to 47.8% at firms with 500 or more lawyers, compared with 17.7% for solo practitioners. Accuracy was the top concern, cited by 74.7% of respondents.
Use is broad across the profession. Clio's 2024 Legal Trends Report found that AI adoption among legal professionals rose from 19% in 2023 to 79% in 2024, according to LawNext.
Enterprise tools are being built into workflows. Harvey's March 2026 funding announcement said customers run more than 25,000 custom agents on its platform, and named firms including Ashurst and Baker Donelson as deploying agents.

None of these sources separates transactional associates from other lawyers. It is reasonable to infer that associates at large firms have access to approved AI tools. What they use those tools for, and how often, is not publicly measured.

Time savings: projected, not measured, and not yet passed on

Thomson Reuters' Future of Professionals report (June 2025; 2,275 professionals surveyed in February and March 2025) found that respondents expect AI to save about 5 hours per week within the next year, up from a 4-hour prediction in 2024. These are self-reported expectations across legal, tax and other professions. They are not time-sheet measurements of associates.

Client-side data suggests those savings have not yet reached clients. In the ACC/Everlaw 2025 survey of 657 in-house professionals, only 24% were satisfied with outside counsel's AI adoption for cost-effectiveness, and 61% planned to push for changes in how legal services are priced. An ACC analysis of the same survey found that, among respondents whose firms use generative AI, 59% had seen no clear savings.

Analysis: One plausible explanation, though no public evidence confirms it, is that time saved on first drafts is being absorbed into more volume or more review rather than fewer billed hours. This matters for billing. ABA Formal Opinion 512 (29 July 2024) says lawyers may bill for the time actually spent, including reviewing AI-generated drafts "for accuracy and completeness," but may not charge a client for learning how to use an AI tool. That is according to the ABA's announcement of the opinion.

Where supervision has demonstrably failed

The published record of AI errors comes mostly from litigation, because court filings are public and deal documents are not. These cases still show how a review chain breaks inside large firms, and the same mechanics apply to transactional work.

Lacey v. State Farm (C.D. Cal.). On 6 May 2025, Special Master Michael R. Wilner (ret.) imposed $31,100 in sanctions, jointly and severally, on Ellis George Cipollone and K&L Gates. Plaintiff's brief cited non-existent or inaccurate authorities. According to an analysis by the French firm DDG, sworn declarations showed that one attorney used CoCounsel, Westlaw Precision Drafting and Google Gemini to prepare a draft. The draft was then circulated and filed without review by supervising attorneys, several of whom did not know it was AI-generated. A revised brief still contained errors. The Special Master found the conduct "tantamount to bad faith."

Concord Music Group v. Anthropic (N.D. Cal.). In May 2025, Magistrate Judge Susan van Keulen struck a paragraph of an expert declaration after a citation formatted with Claude came back with an inaccurate title and incorrect authors. The judge noted that a "manual citation check" had been performed but "did not catch" the error, as reported by The Volokh Conspiracy. Latham & Watkins' team had reviewed the declaration.

Mata v. Avianca (S.D.N.Y.). Judge P. Kevin Castel imposed a $5,000 sanction on 22 June 2023 over ChatGPT-fabricated citations, as reported by Bloomberg Law.

Analysis: The common failure in these cases is that each person assumed someone else had verified the work. The person who drafted it relied on the tool, and the reviewer relied on the person who drafted it. In transactional practice the equivalent risks include defined terms carried in from the wrong precedent, cross-references to schedules that don't exist, and jurisdiction-specific representations copied from generic forms. These errors tend to surface in markups from the other side rather than in court. They are less visible, but not less costly.

The rules already require a supervision framework

ABA Formal Opinion 512 applies the Model Rules' duties of competence, confidentiality, communication, candor, supervision and reasonable fees to generative AI. Supervising lawyers have responsibilities under Rules 5.1 and 5.3 for the work of associates and non-lawyers, and that includes work produced with AI. The Lacey order made the same point from the bench: delegating research to an AI tool does not satisfy a signing lawyer's duty of reasonable inquiry.

The "shadow stack" question

There is no public data on how many associates use AI tools their firm hasn't approved. The general workplace data is striking, however. Microsoft and LinkedIn's 2024 Work Trend Index (31,000 people in 31 countries, 8 May 2024) found that 75% of knowledge workers use AI at work, and 78% of those users bring their own AI tools. Law firms are not exempt from that behaviour. Under Opinion 512, entering client information into a tool without appropriate safeguards raises confidentiality issues under Rule 1.6.

Recommendations for managing partners and legal ops

Practical recommendations, framed as analysis.

Give AI-assisted drafts a separate review protocol. Require the drafter to flag which sections were AI-generated, and have reviewers check defined terms, cross-references and jurisdiction-specific provisions against the source documents rather than the AI's summary.
Make verification a named step with a named person, so it cannot be assumed to have happened.
Measure before you price. Track time on AI-assisted tasks at matter level, so the firm can answer client questions about efficiency with its own data.
Reduce the pull of a shadow stack. Make approved tools easy to use for everyday drafting and communications. Publish a short list of what may and may not be entered into unapproved tools.
Train associates on verification, not just prompting. The court episodes above show the gap is in checking the output, not in generating it.

Sources

American Bar Association, "2024 Artificial Intelligence TechReport" (2025): https://www.americanbar.org/groups/law_practice/resources/tech-report/2024/2024-artificial-intelligence-techreport/
LawNext, "AI Adoption by Legal Professionals Jumps from 19% to 79% in One Year, Clio Study Finds" (7 Oct 2024): https://www.lawnext.com/2024/10/ai-adoption-by-legal-professionals-jumps-from-19-to-79-in-one-year-clio-study-finds
Harvey, "Harvey raises growth round at $11 billion valuation co-led by GIC and Sequoia" (25 Mar 2026): https://www.harvey.ai/en-US/blog/harvey-raises-growth-round-at-dollar11-billion-valuation-co-led-by-gic-and-sequoia
Thomson Reuters, "The AI adoption reality check: Firms with AI strategies are twice as likely to see AI-driven revenue growth" (26 Jun 2025): https://www.thomsonreuters.com/en/press-releases/2025/june/the-ai-adoption-reality-check-firms-with-ai-strategies-are-twice-as-likely-to-see-ai-driven-revenue-growth-those-without-risk-falling-behind
Business Wire / ACC and Everlaw, "64% of In-House Counsel Expect GenAI to Reduce Reliance on Outside Counsel, 50% Expect Lower Outside Counsel Costs, ACC/Everlaw Survey Finds" (14 Oct 2025): https://www.businesswire.com/news/home/20251014382116/en/64-of-In-House-Counsel-Expect-GenAI-to-Reduce-Reliance-on-Outside-Counsel-50-Expect-Lower-Outside-Counsel-Costs-ACCEverlaw-Survey-Finds
ACC Docket, "The Transparency Gap: In-house Counsel Lack Clarity on Law Firms' AI Use" (5 Nov 2025): https://docket.acc.com/node/4313
American Bar Association, "ABA issues first ethics guidance on a lawyer's use of AI tools" (29 Jul 2024): https://www.americanbar.org/news/abanews/aba-news-archives/2024/07/aba-issues-first-ethics-guidance-ai-tools/
DDG, "Sanctions imposed for improper use of generative artificial intelligence in federal civil litigation: analysis of the order issued in Lacey v. State Farm" (2025): https://en.ddg.fr/actualite/sanctions-imposed-for-improper-use-of-generative-artificial-intelligence-in-federal-civil-litigation-analysis-of-the-order-issued-in-lacey-v-state-farm
The Volokh Conspiracy (Reason), "Judge Strikes Part of Anthropic (Claude AI) Expert's Declaration, Because of Uncaught AI Hallucination in Part of Citation" (26 May 2025): https://reason.com/volokh/2025/05/26/judge-strikes-part-of-anthropic-claude-ai-experts-declaration-because-of-uncaught-ai-hallucination-in-part-of-citation/
Bloomberg Law, "Phony ChatGPT Brief Leads to $5,000 Fine for NY Lawyers" (22 Jun 2023): https://news.bloomberglaw.com/esg/chatgpt-phony-legal-filing-case-gets-lawyers-a-5-000-fine
Microsoft, "Microsoft and LinkedIn release the 2024 Work Trend Index on the state of AI at work" (8 May 2024): https://news.microsoft.com/2024/05/08/microsoft-and-linkedin-release-the-2024-work-trend-index-on-the-state-of-ai-at-work/