THE TECHNOLOGY BLIND SPOT

On April 28, 2026, in a federal courtroom in San Francisco, a managing partner who never drafted the brief, never signed the brief, and never read the brief learned that his name on the pleading was enough to make the brief his.

Lenden Webb runs a litigation firm in Fresno. The firm bears his name. His associate, Katherine Cervantes, had filed a Joint Discovery Letter Brief in Hill v. Workday nine months earlier. Cervantes drafted it. Cervantes signed it. Cervantes used Westlaw’s CoCounsel AI tool to research it. The brief cited a case that did not exist, South Pointe Wholesale, Inc. v. Vilardi, and misrepresented the holding of one that did.

Webb told the court he had read Thomson Reuters’ accuracy disclaimer on CoCounsel. He had also been told, by the Westlaw sales representative, that there was “no way” the program could produce hallucinations.

Magistrate Judge Peter H. Kang opened his order with a sentence every supervising attorney in the country should read aloud: “At minimum, a supervising lawyer should read and understand the content of all pleadings and check citations to ensure their accuracy.”

The court fined Webb $1,001 personally, payable by him and not by the firm, by May 22, 2026. It ordered four hours of live in-person continuing legal education within six months: two on supervising junior attorneys, two on ethical use of AI. It required Webb to circulate the order, the original Order to Show Cause, the Further Order to Show Cause, and his own responses to every attorney and paralegal in the firm. And it required him to distribute the CLE materials to attorneys, paralegals, and law clerks after he completed the courses.

Kang did not pick $1,001 by accident. Under California Business and Professions Code § 6086.7(a)(3), a court must notify the State Bar of any judicial sanctions against an attorney except discovery penalties and “monetary sanctions of less than one thousand dollars.” Kang penalized Webb $1 above the line where that exception stops. The arithmetic delivers what the order does not say in words.

The Doctrinal Increment

Penalties for AI-fabricated citations are not new. Damien Charlotin’s database catalogs more than 1,300 such decisions worldwide as of May 2026. The pattern is established. An attorney files a brief generated in part by a large language model. A citation does not exist or does not say what the brief claims. The court fines the drafter, refers them to the bar, and publishes a warning.

Webb is different. Webb did not draft the brief. He was not at the keyboard. He never inserted the citation. The court found, on the record, that Webb had no direct involvement in the research, the drafting, or the filing.

What Webb had was his name on the signature block as counsel of record.

The court’s holding, on page 2 of the order, cites the Eastern District of Oklahoma’s October 2025 ruling in Mattox v. Product Innovations Research, LLC: “Supervisors who do not draft or sign submissions to the court, but who remain counsel of record and fail to question the accuracy of defective pleadings, fail in their duty of supervision.”

That is the doctrinal increment. The principle that supervisors carry accountability is older than generative AI. The application of that principle to attorneys who never touched the file is what is new. Counsel of record is now a personal certification that someone other than the drafter has read the document and verified its claims. The certification cannot be delegated. It cannot be presumed. It cannot be inferred from a firm policy.

The Trend, Not the Outlier

Webb is the third decision in seven months to apply this principle to AI-fabricated citations. Johnson v. Dunn, 792 F. Supp. 3d 1241 (N.D. Ala. 2025), penalized three Butler Snow attorneys including practice group leader William Lunsford, whose name appeared on the signature block of a brief containing five hallucinated citations. The Northern District of Alabama disqualified the attorneys, ordered them to notify clients and presiding judges in all matters, and directed them to send the order to the state bar.

Three months later, Mattox v. Product Innovations Research, LLC, 807 F. Supp. 3d 1341 (E.D. Okla. 2025), applied the same logic in the Eastern District of Oklahoma to three supervising attorneys including local counsel whose name on the docket gave the pleadings a “presumption of legitimacy that they did not deserve.”

Six months after Mattox, Webb extends the line: the supervising attorney who neither drafted nor signed.

Three federal districts. Three separate orders. One principle. The trend is no longer experimental.

The Steelman

Webb did almost everything a thoughtful firm leader is told to do.

He paid for training. He flew ten staff members to the ClioCon legal-tech conference in Boston. He attended seminars on AI ethics alongside Cervantes. He purchased Clearbrief, a commercial citation-checking tool, after the first Order to Show Cause issued. He instructed staff to read every cited case before incorporating it into a brief. He held an “all hands” meeting at the firm’s San Diego office. He wrote a declaration acknowledging “ultimate responsibility as counsel of record.” The court credited every one of these steps. It described his posture more favorably than the supervising lawyer’s in Johnson v. Dunn.

The court still penalized him.

On page 11 of the order, the court explains the reason: “Attorney Webb’s declarations do not describe new or specific procedures for he himself to check citations in briefs which are filed under his name as counsel of record.”

The Webb order is not a punishment for indifference. It is a punishment for a category error. Training is firm-wide. Policy is firm-wide. Procurement of citation-checking software is firm-wide. None of it is the personal pre-filing review that Rule 5.1, and now the federal magistrate’s explicit reading of it, requires of the supervising attorney whose name appears on the document.

Webb’s CLE-and-policy package addressed how the firm functions. The court asked how he functions when a brief leaves the office under his name. He could not answer.

The Real Punishment

The $1,001 fine is the headline. The actual punishment is what follows.

Webb must circulate the order, both prior Orders to Show Cause, and his own declarations to every attorney and paralegal in the firm. He must file a sworn certification with the court that he has done so. After he completes the CLE, he must distribute the educational materials to all attorneys, paralegals, and law clerks. He must file another sworn certification.

The arithmetic of that requirement is worth pausing on. A managing partner has been ordered, by a federal judge, to hand-deliver evidence of his own professional failure to the people he supervises. Not redacted. Not summarized. The full order, with the named-attorney finding that he failed in his duty of supervision, distributed under his own signature to the staff he asks to follow his lead.

Cervantes has worked with Webb for seven years. She will receive a copy. So will the associates Webb hired after her. So will the paralegals. So will the law clerks rotating through the firm this summer. Each of them will read the court’s finding that the partner whose name they work under did not read the brief that bore his signature line.

The fine deters. The circulation rebuilds the supervisory structure from the inside, on the firm’s own letterhead, in a hand the firm cannot control.

What California Is About to Codify

Webb’s order is not the end of this trajectory. It is the leading edge.

On March 13, 2026, the State Bar of California’s Standing Committee on Professional Responsibility and Conduct approved proposed amendments to six Rules of Professional Conduct addressing AI use: Rules 1.1, 1.4, 1.6, 3.3, 5.1, and 5.3. The 45-day comment period closed May 4. The Board of Trustees votes on related practical guidance on May 14 and 15.

The proposed amendment to Rule 5.1 would clarify that managerial lawyers must make reasonable efforts to establish internal policies and procedures governing the use of artificial intelligence. That language, if adopted, will sit alongside Kang’s reading of California’s existing Rule 5.1, the rule he applied to Webb. The court enforced today what the rule will formalize next week.

For attorneys outside California, the question is not whether this language reaches their state. Forty-two jurisdictions have adopted some version of Model Rule 1.1, Comment 8. Rule 5.1 exists in every state’s professional conduct framework. California is the first to name AI governance explicitly. It will not be the last.

Where the Junior-Associate Analogy Breaks

The standard reframe, “supervise AI output the way you supervise junior associate output,” captures the obligation but understates the exposure.

A junior associate carries an independent bar license, a personal duty under Rule 1.1, and individual exposure to penalties when the work product fails. A supervising attorney delegates to a person whose error has its own consequences and its own deterrent. The chain of accountability is layered.

Large language models have none of those features. They carry no license. They owe no duty. They bear no consequences for fabrication. The only deterrent in the chain is the lawyer who reviews the output before it leaves the office. When that lawyer is not the drafter, the only deterrent left is the supervising attorney who is counsel of record. Webb’s name on the brief was not redundant supervision. It was the only supervision left.

That asymmetry sharpens, rather than parallels, the supervisory duty when AI produces the underlying research. The “verify like a junior associate” framing is a floor, not a ceiling. The Webb order treats it accordingly. [See When Attorneys Stop Checking AI Work, The Technology Blind Spot (2026).]

The Conversation Catherine Has on Thursday

The action this case demands is concrete enough to execute before close of business Friday.

Pull every brief, motion, and letter filed in the last 90 days where a managing partner’s name appears as counsel of record but where someone else drafted the document. For each, identify who verified the citations against primary sources, when that verification occurred, and where the verification is documented. The documentation can be a checklist, a written sign-off, an email chain, or an entry in a matter management system. What it cannot be is “the associate confirmed it” or “we use Clearbrief now.”

If the answer to “where is the verification documented” is silence on any filing, the firm has a Rule 5.1 problem that no AI policy resolves. The Webb order describes the remedy: a written, personal pre-filing review procedure that names the supervising attorney who personally checked each citation in each document filed under their name.

The procedure does not need to be elegant. It needs to be specific, written, and used. Webb’s failure was not the absence of training. It was the absence of evidence that he read the brief that bore his name. [See The Yes Machine Problem, The Technology Blind Spot (2026).]

The Apology Is Now Precedent

At the November 4 hearing, Webb said something the court included in its order: “If my name’s on the pleading, even if I didn’t do the research, I should be more familiar with it, and I apologize.”

The court accepted the apology. It did not waive the consequence.

That sentence is now the floor for every supervising attorney whose name appears as counsel of record on a brief researched, drafted, or assisted by a tool that does not know what it is writing. Webb apologized for the position he was in. The next attorney in his position will not have that apology heard for the first time.

The Westlaw sales representative said there was “no way” the tool could produce hallucinations. Twenty-five days before Kang issued his order, the Sixth Circuit had removed a federal criminal appeals attorney for filing a brief generated with the same product. The fabricated quotations in United States v. Farris came from CoCounsel. [See Your AI Research Tool Fabricated the Quotation, The Technology Blind Spot (2026).] Webb believed the representative. The court did not.

By Friday, every managing partner reading this will know whether their firm’s pre-filing review procedure can survive the same hearing. Or whether the only thing standing between their name and a $1,001 personal penalty is the assumption Webb made: that someone else read the brief.

This blog provides general information for educational purposes only and does not constitute legal advice. Consult qualified counsel for advice on specific situations.

About the Author

JD Morris is Co-Founder and COO of LexAxiom, an Agentic AI platform for the business of law. Over a 25-year career, he has built and scaled enterprise technology products across Dell, EMC, VMware, and Cisco, including the first exabyte eDiscovery platform. He holds dual MBAs from Columbia Business School (Finance) and UC Berkeley Haas (Marketing), a Master of Legal Studies in Cybersecurity Law from Texas A&M, and a Master of Engineering from George Washington University. He writes The Technology Blind Spot on the intersection of emerging technology and law. Connect with him on LinkedIn at www.linkedin.com/in/jdavidmorris, on X at @JDMorris_LTech, or on Bluesky at @JDMorris-ltech.bsky.social.

References

1.          Hill v. Workday, Inc., No. 23-cv-06558-PHK, ECF No. 230 (N.D. Cal. Apr. 28, 2026) (Kang, M.J.).

2.         Mattox v. Product Innovations Research, LLC, 807 F. Supp. 3d 1341, 1351 (E.D. Okla. 2025) (Robertson, M.J.).

3.          Johnson v. Dunn, 792 F. Supp. 3d 1241 (N.D. Ala. 2025).

4.         United States v. Farris, No. 25-5623, 2026 WL 915082 (6th Cir. Apr. 3, 2026) (per curiam).

5.          Cal. Bus. & Prof. Code § 6086.7 (West 2025).

6.         Cal. Rules of Pro. Conduct r. 5.1 (Cal. State Bar 2018).

7.          Model Rules of Pro. Conduct r. 5.1 (Am. Bar Ass’n 2024).

8.         Model Rules of Pro. Conduct r. 1.1 (Am. Bar Ass’n 2024).

9.         State Bar of Cal., Standing Comm. on Pro. Resp. & Conduct, Proposed Amendments to the Rules of Professional Conduct Related to Artificial Intelligence (Mar. 13, 2026), https://www.calbar.ca.gov/public/public-meetings-comment/public-comment/public-comment-archives/2026-public-comment/proposed-amendments-rules-professional-conduct-related-artificial-intelligence.

10.    Sam Skolnik, Lawyer in AI Misuse Case Fined Over Subordinate Supervision, Bloomberg L. (Apr. 29, 2026), https://news.bloomberglaw.com/litigation/lawyer-in-ai-misuse-case-fined-for-not-supervising-subordinate.

11.     Damien Charlotin, AI Hallucination Cases (database) (last visited May 7, 2026), https://www.damiencharlotin.com/hallucinations/.