Windows Errors? Fix Them Before They Spread
Repair common Windows errors and clear accumulated junk for a smoother, more stable PC - no reinstall needed.Free scan · no reinstallOutdated Drivers Are Slowing You Down
One free scan finds every outdated or missing driver and matches the right update for your exact hardware.Free scan · exact hardware matchYes, the incident was real—but ChatGPT did not charge a lawyer $10,000. In a 2025 California appeal, the Court of Appeal ordered attorney Amir Mostafavi to pay a $10,000 court sanction after finding that his brief contained fabricated legal quotations and citations generated with help from multiple AI tools. The court said 21 of 23 quotations in the opening brief were fabricated. Mostafavi had not verified them against the underlying cases.
What happened in the California case
The case was Noland v. Land of the Free, L.P., an appeal arising from an employment-related dispute. Mostafavi represented the plaintiff and filed appellate briefs challenging a summary judgment ruling. The California Court of Appeal, Second District, found that the briefs relied on fabricated quotations and citations, as well as authorities that did not support the propositions for which they were cited. The published opinion recounts the court’s findings and the resulting orders.
After the court issued an order requiring him to explain why sanctions should not be imposed, Mostafavi acknowledged using ChatGPT and other generative-AI systems. He said he drafted the briefs and then used AI to “enhance” and check them, but had not manually verified the quotations. The tools he acknowledged included ChatGPT, Claude, Gemini and Grok.
The court affirmed the judgment below and found the appeal frivolous. It ordered Mostafavi to pay $10,000 to the court’s general fund, serve the opinion on his client and certify that he had done so. The clerk was also directed to forward the opinion to the State Bar. The opinion was issued in September 2025; this is a report about a past decision, not a newly imposed penalty.
Were all the citations fake?
No. The often-repeated figure is that 21 of 23 quotations from cases in the opening brief were fabricated—not that every citation in the filing was fake. The court also found that many cited cases did not address the propositions attributed to them and that some authorities did not exist. Mostafavi argued that most citations were accurate, but the court concluded that valid citations elsewhere did not cure the fabricated or unsupported material. CalMatters’ contemporaneous report likewise describes the finding as concerning the quotations.
Why a fabricated legal citation can look real
A legal AI hallucination is a plausible-looking but false or misleading output. It can take several forms:
- A case that does not exist.
- A real case paired with the wrong citation, court or year.
- A quotation that does not appear in the cited opinion, or language assigned to the wrong case.
- A real decision that does not support the legal proposition attached to it.
Formal case names, reporter citations and confident legal prose can make a false reference look credible. An editing request can also create substantive risk: asking a system to improve or polish a brief does not ensure that it will change only style. It may introduce new claims, quotations or citations.
Why the lawyer—not the chatbot—was sanctioned
The court did not impose a fee for using ChatGPT, and the $10,000 was not damages payable to OpenAI. It was a judicial sanction for the lawyer’s conduct in filing the briefs. The court pointed to fabricated authority, unsupported legal propositions, violations of California appellate rules and the frivolous appeal. It also said the court had to spend substantial time tracking down nonexistent authorities and researching the issues without reliable assistance from counsel.
Do these 3 things before closing this tab:
1Scan for outdated or missing drivers - takes under a minute2Clear out junk files and repair common Windows errors3Fix the driver behind crashes, sound loss and screen glitchesRank #3
The court’s central point was responsibility: the attorney, not the court or the AI system, had to identify and verify the authorities submitted in a filing. Its warning was not a blanket ban on AI. Rather, a lawyer should not file citations supplied by generative AI—or any other source—without personally reading and verifying them. The opinion was certified for publication, so it is citable and has significance beyond the parties to this dispute.
Consequences beyond the $10,000 sanction
The direct consequences stated in the opinion were payment to the court’s general fund, client notification and transmission of the opinion to the State Bar. The State Bar referral is not itself a finding of additional discipline; the cited sources do not establish that Mostafavi was disbarred or that the referral resulted in a separate sanction. The case also carries reputational consequences, but the opinion does not establish that the AI-generated citations alone caused the client’s underlying loss.
Rank #4
How this differs from Mata v. Avianca
Noland is separate from the 2023 federal case Mata v. Avianca, in which a judge sanctioned lawyers after they submitted a brief containing nonexistent authorities generated by ChatGPT. That case involved a $5,000 sanction. Noland was a California state appellate proceeding involving a $10,000 sanction and multiple AI tools. The California opinion discusses the broader history of courts confronting fabricated AI-generated authorities; the cases should not be treated as one incident.
Nor should the $10,000 figure be called a permanent “largest-ever” record. Contemporaneous reporting described it as apparently the largest California state-court sanction for attorney AI fabrication at the time. A separate 2025 federal case, Lacey v. State Farm General Insurance Co., involved approximately $31,100 in fees and costs related to bogus AI-generated research by two law firms, a different kind of order that should not be compared as though it were the same fine. The Lacey order sets out that case’s disposition.
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
How lawyers can verify AI-assisted legal work
AI can help organize arguments or improve prose, but citations and quotations should be treated as unverified until checked against authoritative sources. A practical review should include:
- Search the case name and citation in a trusted legal database or the issuing court’s records.
- Open the actual opinion; do not rely on a search-result snippet or an AI-generated summary.
- Confirm the case exists and verify its court, date and citation.
- Compare every quotation with the opinion, including its context and pin cite.
- Check that the holding supports the precise proposition in the brief and that the case remains good law.
- Read every authority in the final filing and keep a research trail showing how it was checked.
- After any AI editing pass, review the whole document again for new or altered factual and legal assertions.
If a false citation is discovered before filing, remove the unsupported point or replace it with verified authority. Do not rely on the same general-purpose chatbot to validate its earlier output; check independently. If the document has already been filed, promptly consult ethics counsel about whether correction, disclosure, withdrawal or another step is appropriate. The right response depends on jurisdiction, court rules and the case’s posture; concealing the error or filing more AI-generated material to explain it can make matters worse.
Choosing legal research tools without outsourcing judgment
General-purpose chatbots can be useful for drafting, rewriting, summarizing and organizing material supplied by a lawyer. They are a poor substitute for authoritative legal research or unsupervised citation generation: they can invent authorities, repeat a user’s incorrect citation, and may not provide a dependable source trail. Use of public or consumer-facing services can also raise confidentiality and data-governance questions depending on the product, settings, firm policy and jurisdiction.
Purpose-built legal research products, including Westlaw Precision, Lexis+ AI, CoCounsel and vLex’s AI tools, are designed around legal content or professional workflows and may make it easier to trace answers to sources. They do not eliminate errors or the need to read the underlying authority. A study testing legal AI research systems reported hallucination rates of roughly 17% to 33% under its particular test conditions; that result is not a universal current accuracy rate because products, versions, prompts and methods differ. The study’s abstract and paper describe its evaluation.
Free tools Windows power users keep installed
One-click scans. No signup required.
For California cases, the Judicial Branch provides a free collection of published and citable appellate opinions. It can help confirm a case and read the opinion, though its coverage and search and citator features differ from commercial research services. California’s opinions page is a primary-source starting point. Whether using a commercial database or a free court source, the useful standard is traceability: locate the authority, read it, check its history and confirm that it supports the filing.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
