technicalmarketing.ai Free AI audit

Report

What is the risk of AI mistakes for a law firm?

The risk of AI mistakes at a law firm has two parts: how often legal AI is wrong and what one error costs. Research tools hallucinated in 17% to 33% of answers in one test1. Courts ordered $5,0002 and $3,0003 over fabricated citations, and a tribunal ordered $812.02 over a chatbot4. The rate for a firm's front desk is NF.

Where does AI touch a law firm, and what does a mistake look like?

A phone agent answers when no one at the desk can. Its mistake is a statement the firm now owns, such as a price, an opening hour or a promise, or an answer that sounds like legal advice. In Moffatt v. Air Canada the Civil Resolution Tribunal found the airline liable for negligent misrepresentation after its website chatbot gave a customer wrong information4. The tribunal wrote that the airline is responsible for all the information on its website, whether it comes from a static page or a chatbot4. That is one tribunal and not a case about lawyers, so for a firm it is an analogy and not a holding.

A second phone mistake makes no noise. A person who consults a lawyer about a possible engagement is a prospective client under Model Rule 1.18, and the lawyer may not use or reveal what the person said even when no relationship follows5. Florida Ethics Opinion 24-1 says a website chatbot used for client intake presents extra risk, including that a prospective-client or even a lawyer-client relationship may arise without the lawyer's knowledge6. Whether a telephone voice agent is a chatbot in that opinion's sense is a question for counsel.

A site assistant carries the same risks in text. California's 2026 guidance, approved May 14, 2026 to replace the 2023 version7, speaks to systems that act on their own. It says attorneys must not deploy agentic systems in a way that lets them make substantive legal determinations, communicate legal advice, prepare and file pleadings, or act in a representative capacity without meaningful attorney supervision and review7. Florida says a chatbot that communicates with clients or third parties must comply with the advertising restrictions and must say it is an AI program and not a lawyer or employee of the firm6.

Marketing copy is the third point. Model Rule 7.1 bars a false or misleading communication about the lawyer or the lawyer's services, including one that omits a fact needed to keep it from misleading8. Comment [3] says an unsubstantiated claim or comparison may mislead when it is stated with enough specificity that a reasonable person would think it can be substantiated, and that a truthful report of past results may create an unjustified expectation for other clients9. A tool that drafts a page for every practice area and city multiplies the claims that need support. A drafted reply to a negative review is a separate trap. ABA Formal Opinion 496, dated January 13, 2021, says a negative online review alone does not meet the requirements for disclosing client information in self-defense under Rule 1.6(b)(5)10. A tool that writes reviews creates a federal problem. Under 16 CFR 465.2(a) it is a deceptive practice for a business to write, create or sell a consumer review that materially misrepresents that the reviewer exists, used the service, or what the experience was11. The FTC said the rule addresses AI-generated fake reviews12.

The lawyer's own work is where the sanctions are. In Wadsworth v. Walmart, motions in limine cited nine cases, of which eight did not exist3. In Mata v. Avianca the court found the respondents abandoned their responsibilities by submitting opinions and quotations that ChatGPT created, and then stood by them after judicial orders questioned their existence2. The same order said there is nothing inherently improper about using a reliable AI tool, and that the rules give attorneys a gatekeeping role over the accuracy of their filings2. A different mistake leaves no trace in a filing. ABA Formal Opinion 512 says a client's informed consent is required before information relating to the representation goes into a self-learning generative AI tool13.

We are not lawyers. Read the rule yourself or ask your counsel how it applies to you.

Risk here is a rate times a consequence. The rate comes from tests, which measure other tools on other tasks, and the consequence comes from orders and statutes. The tests are below, and none is a current error rate.

Dahl and co-authors report in a preprint that legal hallucinations occurred between 58% of the time with ChatGPT 4 and 88% with Llama 2, on specific, verifiable questions about random federal court cases14. Those are general-purpose models and not products sold to law firms. Magesh and co-authors tested the research tools from LexisNexis and Thomson Reuters and found each hallucinated between 17% and 33% of the time, less than general-purpose GPT-4 but contrary to the providers' claims of being hallucination-free1. In the same Stanford evaluation Lexis+ AI answered 65% of queries accurately15, and Westlaw's AI-Assisted Research was accurate 42% of the time and hallucinated nearly twice as often as the other legal tools tested15. Accurate and hallucinated are separate measures, so the figures do not add to 100%. The benchmark held 202 queries, which its authors call small15.

A third test came out differently. Vals AI says that in its October 2025 benchmark of 200 legal research questions all the AI products scored 74% to 78% on an average weighted score, against 69% for a lawyer baseline16. The products, questions and scoring differ from the Stanford test, so we do not average the two or pick one as the rate.

Counts of sanctions are not rates either. Charlotin's AI Hallucination Cases database, a running count by one researcher, listed 2,289 decisions where a court found or implied that a party relied on hallucinated content when we read it on November 10, 202617. It does not say how many filings were checked.

The nearest measurement to a voice agent on a front desk comes from a hospital. At Mount Sinai an AI voice assistant making pre-procedure calls completed 86.4% of calls in Phase I18, and an AI hallucination led to an incomplete call in 6 calls, or 0.7%18. It counted only hallucinations that ended a call, and its calls went out to scheduled patients on a script. A field study of an AI agent answering inbound calls at a law firm is NF, and so is a measure of the time a lawyer spends checking AI output. The nearest evidence is the research tests above.

The other side of the ledger is the call nobody answers. Clio† (practice-management software) says that in its 2024 secret-shopper study only 40% of 500 law firms answered the phone, and 52% either answered or called back19. This page prices the mistake side only.

What does one AI mistake cost a law firm?

The consequences that sources price are in the table. Amounts are as the sources state them.

Kind of errorSourceAmount as the source states itCaveat
Fabricated citations filedMata v. Avianca, S.D.N.Y., June 22, 2023$5,000 penalty, jointly and severally on two lawyers and their firm2One order
Fabricated citations filedWadsworth v. Walmart, D. Wyo., order of February 24, 2025$3,000 fine and revoked admission for one attorney; $1,000 each for two others3One order
Fabricated citations filedJohnson v. Dunn, N.D. Ala., July 23, 2025Public reprimand, disqualification, published order, bar referral20No money amount in our evidence
Chatbot misstates a policyMoffatt v. Air Canada, 2024 BCCRT 149$812.02 in total4A Canadian tribunal, not a lawyer case
Call recorded without all-party consentCalifornia Penal Code 632(a)Fine up to $2,500 per violation, up to $10,000 after certain prior convictions21Confidential communications only; Washington's penalty is NF
Call placed with an AI voice without consent47 U.S.C. 227(b)(3)Actual loss or $500 per violation, whichever is greater; up to 3 times that if willful or knowing22Calls the firm places; the FCC's statement on inbound calls is a proposal23
Fake or AI-generated review16 CFR part 465Civil penalties up to $53,088 per violation24Courts may impose them for knowing violations25
Claim that an AI does a lawyer's workFTC final order, DoNotPay$193,000 and notice to subscribers26A settlement with a seller of AI, not a law firm

A sanction order or a settlement is one case and not a rate. How often a firm of this size is sanctioned, fined or sued over an AI error is NF, and so are the legal fees and the owner's time that no source prices. The row types behave differently. A statutory amount per violation, such as $500 or $2,500, grows with the number of contacts the error touches. A sanction behaves like an incident. We think a scope limit that removes a kind of contact removes the per-contact exposure, and that a review step lowers the chance of an incident.

Does malpractice insurance cover AI mistakes?

The evidence is thin. ALPS, a legal malpractice insurer, wrote on August 21, 2025 that lawyers' professional liability policies typically do not exclude coverage for negligence claims arising from the use of generative AI, though coverage may depend on whether the conduct meets the policy's definition of professional services27. The same post says that if firm lawyers let an AI make critical legal judgments without attorney oversight, that could be viewed as the unauthorized practice of law, which such policies typically exclude27. It is a risk manager's article and not policy wording, and "typically" is the author's generalization across the market.

On the general liability side, a trade-association summary of Verisk's announcement says that from January 2026 ISO endorsements let insurers exclude generative AI exposures from commercial general liability policies28. An insurer chooses whether to use the forms, and general liability is a different policy from professional liability. A state bar statement on malpractice coverage of AI errors is NF; the nearest evidence is the ALPS article. The firm reads its own policy and asks its broker.

How does a law firm lower the cost of an AI mistake?

Five controls act on the cost: scope limits, disclosure, hand-off, logging and review. Each is a decision for the firm, and each has a text behind it.

Scope limits. The firm decides what the agent may state and do. We think the right list is short: practice areas, hours and the fees the firm has published, with booking and message-taking as the actions. We think what stays with a lawyer, whoever or whatever answers, is whether the caller has a case, whether the firm will take it, the state of an existing matter, any fee beyond the published one and whether a conflict exists. California's guidance draws a similar line for systems that act7. OWASP defines Excessive Agency as the vulnerability that lets damaging actions follow unexpected, ambiguous or manipulated model output29. Rule 1.18 argues for collecting little, and we think four items are enough: a way to reach the person, the kind of matter, what the lawyers say a conflict check needs and a booked time5.

Disclosure. The agent's first sentence says what it is. Florida's opinion requires a chatbot disclaimer that it is an AI program and not a lawyer or employee of the firm6. ABA Opinion 512 says a lawyer must disclose AI practices if a client asks how the work was done or whether AI was used, or if the client requires disclosure13. For filings, the Northern District of Texas requires in Local Civil Rule 7.2(f) that a brief prepared using generative AI say so on its first page under the heading "Use of Generative Artificial Intelligence"30, under orders of June 10, 2024, effective September 3, 202431. California's guidance tells an attorney to check for rules or orders in the jurisdiction that require disclosure7. Someone at the firm decides who checks the local rules before each filing.

Hand-off. The firm names the person who receives a transferred call, how fast that person acts and what the agent passes along. We think the summary should say what the caller asked and what the agent answered. The CFPB's 2023 report collected complaints of customers stuck in loops with chatbots, often because the issue was outside the chatbot's limited capabilities32. Urgent matters such as a hearing, an arrest or a filing deadline go to a lawyer the firm names.

Logging. A log shows what the agent said and did, and it is also data. The firm decides what is recorded, who may read it and how long it is kept; we give no retention period. A recording engages the statutes. Federal law lets a party record or consent to a recording by one party33, Washington requires the consent of all participants34, and California punishes a recording made without the consent of all parties to a confidential communication21. The contents of the log are information relating to the representation, and Rule 1.6(c) asks for reasonable efforts to prevent its inadvertent or unauthorized disclosure35. Opinion 512 says boilerplate in an engagement letter is not enough to obtain informed consent13.

Review. A person reads what the AI produced. Opinion 512 says submitting output without appropriate independent verification could violate the duty of competence13. Texas Opinion 705 says a lawyer should always verify the accuracy of a generative AI tool's responses36. California names two duties, technological competence and independent professional judgment in reviewing, verifying and correcting outputs7. Review fails when the reviewer approves without checking. NIST's profile lists automation bias and over-reliance among its human-AI configuration risks37. For marketing, we think a lawyer should read every page, profile and advertisement a vendor drafts before it goes live, and a claims register should record who wrote each claim, what supports it and when it was last checked. Rule 7.2(d) asks that a communication under the rule name at least one responsible lawyer or firm38. Review replies use a template a lawyer approved, with nothing about any client's matter, and a rule about who may post it.

What stays with the lawyer?

The texts put several duties on the lawyer and not on the tool. The lawyer verifies anything filed or sent2,13. The duty of candor to the tribunal cannot be delegated to AI7. The decision to put client information into a tool, and the client's informed consent, stay with the lawyer13. So does what to tell the client about AI use7. The fee also stays: a lawyer billing hourly must bill actual time13, and Texas says not the time that was saved36.

Supervision stays too. Opinion 512 says lawyers have additional supervisory obligations when they rely on others outside the firm to employ AI tools13. Model Rule 5.3 asks a partner or lawyer with comparable authority to make reasonable efforts to ensure the firm has measures giving reasonable assurance that a nonlawyer behaves compatibly with the lawyer's obligations39. Whether a given tool is such a nonlawyer is for counsel. The state versions of these model rules differ, so the firm's counsel reads its own.

Where our service fits

We sell an AI voice agent for $299 a month with 250 call minutes included, then $0.39 a minute40, added to a plan at $999, $1,999 or $2,999 a month per location40. Our design intent is that the agent says it is an AI assistant first, collects the intake fields the firm chose, books a consultation or takes a message, transfers an urgent call to the lawyer the firm names and gives no legal advice, with a lawyer at the firm reviewing the script before it goes live. A measured error rate for our agent is NF, and we promise no outcome.

People also ask

Can a law firm be sanctioned for an AI mistake?

Yes, in the orders we hold. A court ordered $5,000 in Mata v. Avianca2, and in Wadsworth v. Walmart it fined one attorney $3,000, revoked his admission and fined two others $1,000 each3. Each is one order and not a rate.

Is a lawyer responsible for what an AI chatbot tells a client?

The tribunal in Moffatt v. Air Canada held a company responsible for its chatbot's wrong answer4. That was not a lawyer case. For lawyers, Rule 5.3 asks for measures giving reasonable assurance that a nonlawyer behaves compatibly with the lawyer's obligations39, and whether a tool counts is for counsel.

One Stanford test found the research tools from LexisNexis and Thomson Reuters hallucinated between 17% and 33% of the time1. Vals AI says its October 2025 benchmark scored products 74% to 78% against a lawyer baseline of 69%16. The tests differ and neither is a current rate.

Does malpractice insurance cover AI errors?

ALPS, a legal malpractice insurer, wrote that lawyers' professional liability policies typically do not exclude negligence claims arising from generative AI use27. It also wrote that letting AI make critical legal judgments without oversight could be unauthorized practice, which such policies typically exclude27. Read your policy.

Does a law firm have to tell clients it uses AI?

ABA Opinion 512 says disclosure may be unnecessary depending on the circumstances, but a lawyer must disclose AI practices if the client asks or requires it13. California's guidance says an attorney must consider disclosing an intention to use AI7. State rules differ.

Can a law firm put client information into ChatGPT?

ABA Opinion 512 says a client's informed consent is required before information relating to the representation goes into a self-learning generative AI tool, and boilerplate in an engagement letter is not enough13. California says not without informed consent where the tool may present material confidentiality risks7.

Can AI write a law firm's online reviews or replies?

An AI-written review that misrepresents who the reviewer is, or what the experience was, is a deceptive practice under 16 CFR 465.2(a), with penalties up to $53,088 per violation11,24. A reply may not disclose client information merely because a review is negative10.

How to read this page: † a vendor’s own figure, NF not found, ( ) our arithmetic on cited figures, and a small number points to a source below.

Sources

  1. Varun Magesh, Faiz Surani, Matthew Dahl, Mirac Suzgun, Christopher D. Manning, Daniel E. Ho. Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools. arXiv (submitted 30 May 2024), 2024. arxiv.org/abs/2405.20362
  2. Judge P. Kevin Castel, U.S. District Court, S.D.N.Y.. Mata v. Avianca, Inc., No. 22-cv-1461 (PKC), Opinion and Order on Sanctions (Doc. 54). U.S. District Court for the Southern District of New York (RECAP/CourtListener copy of the filed order), 2023-06-22. storage.courtlistener.com/recap/gov.uscourts.nysd.575368/gov.uscourts.
  3. U.S. District Court for the District of Wyoming. Wadsworth v. Walmart Inc., No. 2:23-CV-118-KHR, Order on Sanctions and Other Disciplinary Action (Doc. 181). U.S. District Court for the District of Wyoming (RECAP/CourtListener copy of the filed order), 2025-02-24. storage.courtlistener.com/recap/gov.uscourts.wyd.64014/gov.uscourts.wy
  4. Christopher C. Rivers, Tribunal Member. Moffatt v. Air Canada, 2024 BCCRT 149. Civil Resolution Tribunal of British Columbia, 2024-02-14. decisions.civilresolutionbc.ca/crt/crtd/en/525448/1/document.do
  5. American Bar Association. Rule 1.18: Duties to Prospective Client (ABA Model Rules of Professional Conduct). American Bar Association, 2020. americanbar.org/groups/professional_responsibility/publications/model_
  6. The Florida Bar. Florida Bar Ethics Opinion 24-1 (January 19, 2024). The Florida Bar (copy hosted by University at Buffalo School of Law CLE materials), 2024-01-19. law.buffalo.edu/content/dam/law/content/cle/24mar27-materials/fl-bar-e
  7. State Bar of California, Committee on Professional Responsibility and Conduct. Practical Guidance for the Use of Generative Artificial Intelligence in the Practice of Law (Executive Summary). State Bar of California, 2026-05-14. calbar.ca.gov/Portals/0/documents/ethics/Generative-AI-Practical-Guida
  8. American Bar Association. Rule 7.1: Communications Concerning a Lawyer's Services (ABA Model Rules of Professional Conduct). American Bar Association, 2024. americanbar.org/groups/professional_responsibility/publications/model_
  9. American Bar Association. Comment on Rule 7.1 (ABA Model Rules of Professional Conduct). American Bar Association, 2024. americanbar.org/groups/professional_responsibility/publications/model_
  10. ABA Standing Committee on Ethics and Professional Responsibility. ABA Formal Opinion 496: Responding to Online Criticism. American Bar Association, 2021-01-13. americanbar.org/content/dam/aba/administrative/professional_responsibi
  11. Federal Trade Commission. 16 CFR 465.2 Fake or false consumer reviews, consumer testimonials, or celebrity testimonials. Legal Information Institute, Cornell Law School (Code of Federal Regulations text), 2024-10-21. law.cornell.edu/cfr/text/16/465.2
  12. Federal Trade Commission. Federal Trade Commission Announces Final Rule Banning Fake Reviews and Testimonials. Federal Trade Commission, 2024-08-14. ftc.gov/news-events/news/press-releases/2024/08/federal-trade-commissi
  13. ABA Standing Committee on Ethics and Professional Responsibility. ABA Formal Opinion 512: Generative Artificial Intelligence Tools. American Bar Association, 2024-07-29. americanbar.org/content/dam/aba/administrative/professional_responsibi
  14. Matthew Dahl, Varun Magesh, Mirac Suzgun, Daniel E. Ho. Large Legal Fictions: Profiling Legal Hallucinations in Large Language Models. arXiv (v2, last revised 21 Jun 2024), 2024. arxiv.org/abs/2401.01301
  15. Varun Magesh, Faiz Surani, Matthew Dahl, Mirac Suzgun, Christopher D. Manning, Daniel E. Ho. Hallucination-Free? Assessing the Reliability of Leading AI Legal Research Tools. arXiv (published in the Journal of Empirical Legal Studies, 2025), 2024. arxiv.org/pdf/2405.20362
  16. Vals AI. VLAIR - Legal Research. Vals AI, 2025. vals.ai/industry-reports/vlair-10-14-25 Secondary source.
  17. Damien Charlotin. AI Hallucination Cases Database. Damien Charlotin (personal research database), 2026-11-10. damiencharlotin.com/hallucinations/ Secondary source.
  18. Annapoorna Kini et al.. Conversational artificial intelligence for pre-procedural patient preparation: implementation, validation and patient satisfaction. npj Digital Medicine, 2025. pmc.ncbi.nlm.nih.gov/articles/PMC13710839
  19. Clio. Highlights from the Legal Trends Report: the Legal Industry in 2024. Clio (a legal practice-management software vendor), 2024. clio.com/blog/highlights-from-2024-legal-trends-report/ Secondary source.
  20. U.S. District Court for the Northern District of Alabama. Johnson v. Dunn, No. 2:21-cv-1701-AMM, Sanctions Order (Doc. 204). U.S. District Court for the Northern District of Alabama (Docket Alarm copy of the filed order), 2025-07-23. docketalarm.com/cases/Alabama_Northern_District_Court/2--21-cv-01701/J
  21. California Legislature. California Penal Code section 632. California Legislative Information (leginfo.legislature.ca.gov), 2026. leginfo.legislature.ca.gov/faces/codes_displaySection.xhtml?lawCode=PE
  22. United States Congress. 47 U.S.C. 227: Restrictions on use of telephone equipment. Legal Information Institute, Cornell Law School (US Code text), 2026. law.cornell.edu/uscode/text/47/227
  23. Federal Communications Commission. Implications of Artificial Intelligence Technologies on Protecting Consumers From Unwanted Robocalls and Robotexts (notice of proposed rulemaking), 89 Fed. Reg. 73321. Federal Register / U.S. Government Publishing Office, 2024-09-10. govinfo.gov/content/pkg/FR-2024-09-10/html/2024-19028.htm
  24. Federal Trade Commission. FTC Warns 10 Companies About Possible Violations of the Agency's New Consumer Review Rule. Federal Trade Commission, 2025-12-22. ftc.gov/news-events/news/press-releases/2025/12/ftc-warns-10-companies
  25. Federal Trade Commission. The Consumer Reviews and Testimonials Rule: Questions and Answers. Federal Trade Commission, 2024-10-21. ftc.gov/business-guidance/resources/consumer-reviews-testimonials-rule
  26. Federal Trade Commission. FTC Finalizes Order with DoNotPay (FTC press release page). Federal Trade Commission, 2025-01-16. ftc.gov/node/87474
  27. Mark Bassingthwaighte, Risk Manager. Insurance Coverage Issues for Lawyers in the Era of Generative AI. ALPS Corporation, 2025. alpsinsurance.com/blog/insurance-coverage-issues-for-lawyers-in-the-er
  28. Independent Insurance Agents & Brokers. Verisk to Roll Out New General Liability Exclusions for Generative AI Exposures. Big I (independentagent.com), 2026. independentagent.com/vu_resource/verisk-to-roll-out-new-general-liabil Secondary source.
  29. OWASP GenAI Security Project. LLM06:2025 Excessive Agency. OWASP Foundation, 2025. genai.owasp.org/llmrisk/llm062025-excessive-agency/
  30. U.S. District Court for the Northern District of Texas. Local Civil Rules, N.D. Tex., LR 7.2(f) Disclosure of Use of Generative Artificial Intelligence. U.S. District Court for the Northern District of Texas, 2024-09-03. txnd.uscourts.gov/sites/default/files/documents/CIVRULES.pdf
  31. Chief Judge David C. Godbey, U.S. District Court for the Northern District of Texas. Special Order No. 2-98 (adopting new rules LR 7.2(f) and LCrR 47.2(e)). U.S. District Court for the Northern District of Texas, 2024-09-03. txnd.uscourts.gov/sites/default/files/orders/2-98.pdf
  32. Consumer Financial Protection Bureau. Chatbots in consumer finance. Consumer Financial Protection Bureau, 2023-06-06. consumerfinance.gov/data-research/research-reports/chatbots-in-consume
  33. United States Congress. 18 U.S.C. 2511: Interception and disclosure of wire, oral, or electronic communications prohibited. Legal Information Institute, Cornell Law School (US Code text), 2026. law.cornell.edu/uscode/text/18/2511
  34. Washington State Legislature. RCW 9.73.030: Intercepting, recording, or divulging private communication - Consent required - Exceptions. Washington State Legislature, 2026. app.leg.wa.gov/rcw/default.aspx?cite=9.73.030
  35. American Bar Association. Rule 1.6: Confidentiality of Information (ABA Model Rules of Professional Conduct). American Bar Association, 2024. americanbar.org/groups/professional_responsibility/publications/model_
  36. Professional Ethics Committee for the State Bar of Texas. Opinion 705. Texas Center for Legal Ethics, 2025-02. legalethicstexas.com/resources/opinions/opinion-705/
  37. National Institute of Standards and Technology. Artificial Intelligence Risk Management Framework: Generative Artificial Intelligence Profile, NIST AI 600-1. National Institute of Standards and Technology, 2024-07. nvlpubs.nist.gov/nistpubs/ai/NIST.AI.600-1.pdf
  38. American Bar Association. Rule 7.2: Communications Concerning a Lawyer's Services: Specific Rules (ABA Model Rules). American Bar Association, 2024. americanbar.org/groups/professional_responsibility/publications/model_
  39. American Bar Association. Rule 5.3: Responsibilities Regarding Nonlawyer Assistance (ABA Model Rules of Professional Conduct). American Bar Association, 2024. americanbar.org/groups/professional_responsibility/publications/model_
  40. Technical Marketing. Pricing. Technical Marketing, 2026-11. technicalmarketing.ai/pricing/ Our own data.

Find out what AI says about you.

A free 20-minute call. We ask the assistants your customers' questions and show you where you come up today.

No pitch unless you ask for one.