Morgan & Morgan
Morgan & Morgan lawyers sanctioned after the firm's in-house AI tool invented case citations
Morgan & Morgan launched MX2.law, an in-house AI legal assistant, in November 2024. In January 2025 a lawyer used it to add case law to a court motion, and eight of the nine cases cited did not exist. A federal judge fined the lawyer who drafted the motions $3,000 and two other signing lawyers $1,000 each, and did not sanction the firms.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
eight
Cases cited in the January 22, 2025 motions in limine that did not exist (out of nine cited)1
$3,000
Fine imposed on the attorney who drafted the motions (his admission to the case was also revoked)1
$1,000
Fine imposed on each of the two other signing attorneys1
November 15th, 2024
Date the firm launched MX2.law, per its Chief Transformation Officer2
What it took
Cost, team size and timeline were not disclosed in the sources.
The problem
Law firms must not cite cases that do not exist. Courts expect the lawyer who signs a filing to have checked the law in it. Morgan & Morgan, a large US plaintiffs' law firm, wanted its lawyers to work faster with AI, and built its own tool for that.
What they built
MX2.law, which the firm's Chief Transformation Officer described in a court declaration as a private AI legal assistant launched on November 15th, 2024. It was built to protect client data, including health information, and to connect to the firm's document system. Its features included summarising documents, searching within them and helping to draft text. The firm told users that AI answers can be wrong and required attorneys to review AI content before use, and it ran a webinar, written guides and optional hands-on sessions.
How they did it
In a case against Walmart and Jetson Electric Bikes, attorney Rudwin Ayala drafted motions in limine, which are requests to the judge about what evidence the jury may hear. He uploaded the draft to MX2.law and asked it to add case law from Wyoming federal courts. He did not check the cases it added. Two other lawyers also signed the motions without reviewing them.
What went wrong
The motions, filed on January 22, 2025, cited nine cases, and eight did not exist. The judge issued an order to show cause on February 6, 2025. The plaintiffs withdrew the motions the next day. The lawyers then admitted the cases were made up by an AI platform. The court found that all three signers broke the rule requiring a reasonable inquiry into the law, noted the signs that something was off (missing pinpoint pages, odd citation format), and said the duty to check cannot be handed to a colleague.
Results
On February 24, 2025 the court fined Mr. Ayala $3,000 and removed him from the case. It fined each of the other two attorneys $1,000. It chose not to order the firms to explain themselves, citing the firm's training and the new on-screen reminder that users must verify AI output. The judge called the case the latest reminder not to rely blindly on AI citations.
What to copy
Make verification a required step, and make the tool say so at the point of use. Keep records of tool usage so you can show a court exactly what happened.
Lessons for your project
Treat output from an in-house AI tool exactly like output from a public chatbot. The court held that the signing lawyers' duty to check the law cannot be delegated, whatever the tool.
Training and warnings did not prevent the error. Add hard checks before filing, such as a required second pair of eyes for every citation.
Put a verification prompt inside the tool. After the incident the firm added an acknowledgement that users must independently verify AI output, which the judge counted in the firm's favour.
Respond fast and openly. The court cited the withdrawal of the motions, payment of the other side's fees, candour and new safeguards as reasons for lighter penalties.
Caveats
- The model behind MX2.law is not named in the sources. The firm's executive says only that it uses best-in-class AI model functionality.
- The other side's lawyers had suggested ChatGPT might be the source. The judge found the plaintiffs' attorneys had been honest that they used the firm's tool, and did not find ChatGPT was used.
- The attorney who drafted the motions told the court it was his first time using AI to add case law. The record does not show how widely the firm's lawyers used the tool.
- Fines were small. The significance is the reasoning: signature duties apply to AI-assisted drafting.
- The ABA Journal reports the combined sanction as $5,000; this entry gives the three amounts from the court order.
Sources
- Order on Sanctions and Other Disciplinary Action, Wadsworth v. Walmart Inc., No. 2:23-CV-118-KHR (D. Wyo.)
United States District Court for the District of Wyoming (via CourtListener) · 2025-02-24 · Filing · storage.courtlistener.com
- Declaration of Yathavapilas Ithayakumar (Exhibit 4 to Response to Order to Show Cause)
Morgan & Morgan, filed in the District of Wyoming (via CourtListener) · 2025-02-13 · Filing · storage.courtlistener.com
- No. 42 law firm by head count sanctioned over fake case citations generated by AI
ABA Journal · 2025-02-10 · News · abajournal.com
Summarized in our own words from the sources above. Spotted an error? See corrections.