Is AI actually worth it for a solo or small-firm lawyer?
Yes, with a caveat that matters more than the upside. Clio's 2026 Legal Trends for Solo and Small Law Firms report, published May 4, 2026, found that 71% of solo practitioners and 75% of small firms now use AI to complete legal work. But fewer than 33% of solo and small firms have actually increased revenues with AI, versus nearly 60% of enterprise firms. Part of the gap is pricing, not technology: Clio found 86% of solo firms and 78% of small firms made no pricing changes at all after adopting AI, which means an hourly billing model hands the efficiency gain straight to the client.
The other part of the gap is process. Solo and small-firm lawyers face a specific version of the AI problem: real time pressure, no bench of associates to check work, and — per the 8am 2026 Legal Industry Report, a survey of more than 1,300 legal professionals published March 2026 — a governance gap in which 43% of firms have no formal AI policy and no plans to create one, and only 9% have a written policy that is actually enforced. Big-law coverage of "AI for lawyers" assumes a knowledge-management team and an innovation partner. Solo practitioners have neither. This piece is written for the lawyer doing intake, drafting, billing, and research alone or with one paralegal.
Where does AI actually save small-firm lawyers time?
Lawyers see the clearest AI leverage in five tasks: document review, first-draft generation, client intake triage, billing narrative cleanup, and preliminary legal research. Each has a different risk profile, and conflating them is how firms end up either over-trusting research output or under-using document review, which is the safer, higher-leverage task of the two.
Document review and summarization. Feeding a batch of discovery documents or a contract stack to an AI tool for first-pass summarization is the lowest-risk, highest-leverage use case — the AI isn't generating anything that goes in front of a judge, it's compressing volume a human still reads.
Drafting. Lawyers get fast first drafts of routine motions, demand letters, and contract clauses. Clio's coverage of time-strapped solo and small firms describes demand letters that once took weeks to write now taking days, and long email chains distilled to their actionable points. But every draft needs a substantive edit pass before it leaves the building, not a skim.
Intake and triage. Lawyers using AI-assisted intake forms and initial client screening cut the manual back-and-forth of qualifying a lead before billable time gets spent, similar to the intake automation pattern covered in isonew's guide to AI workflow automation for small business.
Billing narratives. Turning raw time entries into client-ready narrative descriptions is a low-stakes, high-annoyance task AI handles well — the exposure here is inaccuracy in what work is described, not confidentiality, so lawyers still need a glance before it goes on an invoice.
Legal research. The highest-leverage and highest-risk use case simultaneously — research tools can surface relevant case law fast, but this is also where the documented sanctions cases below originated.
Why do lawyers keep getting sanctioned for AI-fabricated citations?
Lawyers get sanctioned because generative AI tools invent case names, docket numbers, and quotes that sound real and pass a skim but don't exist — and the lawyer files the brief without independently verifying every citation against a real database. As of June 9, 2026, the public AI Hallucination Cases database maintained by Damien Charlotin, a research fellow at HEC Paris, tracked 1,598 court decisions involving AI-fabricated citations or content, up from roughly 200 a year earlier. As of April 2026, when the database stood at 1,313 documented proceedings, 496 of them involved licensed attorneys rather than self-represented litigants.
This isn't a hypothetical risk category. It's a named, dated pattern:
- Whiting v. City of Athens (Sixth Circuit, March 13, 2026): A panel of Judges Jane B. Stranch, John K. Bush, and Eric E. Murphy sanctioned Tennessee attorneys Van R. Irion and Russ Egli $15,000 each to the court registry, plus joint responsibility for the appellees' appellate fees and double costs, over briefs containing more than two dozen fake citations and misrepresentations of fact. Worth noting for anyone hoping to blame the tool: the panel asked whether generative AI was used and never received an answer, then sanctioned the fabrications regardless of source — writing that no brief should contain citations, "whether provided by generative AI or any other source," that a lawyer has not personally read and verified. Source: LawSites.
- Couvrette v. Wisnovsky (D. Or., December 2025): Roughly $110,000 in combined fines and fee awards against San Diego attorney Stephen Brigandi and Portland local counsel Tim Murphy, whose briefs included 15 references to nonexistent cases and 8 fabricated quotations. Brigandi was ordered to pay $80,000 in opposing attorney fees plus about $15,000 in fines; Murphy drew $14,000 in additional fees. Judge Mark D. Clarke called the conduct "a notorious outlier in both degree and volume." Source: ABA Journal.
- Coomer v. Lindell (D. Colo., 2025): Two attorneys representing MyPillow CEO Mike Lindell, Christopher Kachouroff and Jennifer DeMaster, filed a brief containing nearly 30 defective citations; Judge Nina Y. Wang fined each attorney $3,000, finding no explanation other than "the use of generative artificial intelligence or gross carelessness." Source: Law&Crime.
- Wadsworth v. Walmart (D. Wyo., 2025): Attorneys at Morgan & Morgan, one of the country's largest plaintiff firms, filed motions in limine citing nine cases in a product-liability matter — eight of which did not exist. The citations came from the firm's own in-house AI platform. Three attorneys were fined $5,000 in total, with the drafting attorney fined $3,000 and removed from the case. Source: Bloomberg Law.
"Attorneys must ensure that they detect and correct hallucinations yielded in their research or the work done by legal professionals they supervise." — Cynthia Feathers, writing for the New York State Bar Association, June 2026. Source: NYSBA.
Penalties have escalated well past fines. On June 3, 2026, the Ninth Circuit suspended attorneys Mike Sethi and William Rounds from practice before that court for six months and fined them $2,500 each over briefs with fabricated citations and invented quotes. On June 9, 2026, the Northern District of Mississippi barred attorneys Kathleen Wilson and Kathryn Williams for two years, with fines of $2,500 and $3,500, after AI drafting and research tools produced fabricated case citations in their filings. And the earliest of these, People v. Crabill in Colorado (November 2023), produced a suspension of one year and one day — 90 days served, the remainder stayed pending two years of probation. What links these cases is not firm size. It is a filing that went out without anyone checking the citations.
What do the ABA and state bars actually require of lawyers using AI?
Lawyers using generative AI must maintain competence in how the tools work, protect client confidentiality when inputting case information, and supervise the output as if a junior associate produced it — not a peer. That standard comes from ABA Formal Opinion 512, issued July 29, 2024, the first formal ABA ethics guidance on lawyers' use of generative AI. It addresses six duties: competence, confidentiality, communication, candor, supervision, and reasonable fees.
Two matter most for a small firm with no compliance department:
Confidentiality. Lawyers are responsible for knowing how a generative AI tool uses the data they put into it, and Opinion 512 recommends obtaining a client's informed consent before entering client confidences into one. Feeding case facts into a consumer-grade tool rather than a vetted, firm-licensed one with a data agreement is precisely where that duty gets tested — and Clio found that 47% of solos and 48% of small firms use consumer-grade tools like ChatGPT or Microsoft Copilot.
Supervision. Opinion 512 puts the obligation on partners and managerial lawyers to establish clear policies on permissible AI use and supervise compliance. In practice, a supervising lawyer is responsible for AI output the same way they would be for a paralegal's draft — and "I didn't know the tool made it up" has not carried the day in any of the sanctions cases above. Every AI-assisted work product needs a named human who checked it before it left the building.
Which tasks need which human check?
| Task | AI leverage | Required human check | Ethics exposure |
|---|---|---|---|
| Document review / summarization | High — compresses volume fast | Spot-check summary against source docs before relying on it | Low, if source docs stay in a vetted tool |
| Drafting (motions, letters, clauses) | High — cuts first-draft time | Full substantive lawyer edit before filing or sending | Medium — competence duty if unedited AI language ships |
| Client intake triage | Medium — speeds qualification | Lawyer confirms conflict check and scope before engagement | Low-medium — confidentiality if intake data leaves a vetted system |
| Billing narrative generation | Medium — saves admin time | Lawyer or biller confirms accuracy of described work | Low — billing-accuracy exposure, not confidentiality |
| Legal research / citation generation | High risk if unverified | Independently verify every citation in a real case database (Westlaw, Lexis, PACER) before filing | Highest — sanctions, fines, suspension, referral to bar |
Lawyers reading this table should notice the pattern: the tasks where AI saves the most time and the tasks where it creates the most exposure are not the same tasks, and a firm-wide policy that treats "using AI" as one category misses that entirely.
What should a solo or small-firm lawyer actually do this quarter?
Solo and small-firm lawyers should write a one-page AI policy this quarter — even a five-bullet list of what's allowed, which tool is vetted, and who signs off before filing — because a firm with no policy has nothing standing between a hallucinated citation and a filed brief. Lawyers should pair every AI-assisted research task with an independent citation check in a real database, treat AI drafts as a first-year associate's work product requiring full review, and keep client-identifying facts out of consumer-grade tools without a data agreement. None of this requires a general counsel or a knowledge-management budget. It requires a written rule and a named human accountable to it — and per Clio, 57% of solos and 55% of small firms still have no written AI policy at all.
For the broader small-business AI picture beyond legal-specific tasks, see isonew's AI for small business hub and the sibling breakdown for AI for accountants, who face a parallel verification problem around AI-generated numbers instead of AI-generated citations.