For the first time in a published opinion, the Colorado Court of Appeals has now warned lawyers and unrepresented litigants that Colorado state courts may sanction them for the (mis)use of generative AI. In Al-Hamim v. Star Hearthstone, LLC, Al-Hamim, representing himself, sued Star Hearthstone, LLC asserting claims of breach of the warranty of habitability. The facts of the underlying dispute are unremarkable and not central to the issue of sanctions for the of generative AI.
Al-Hamim lost in the district court, with his pro se complaint being dismissed for failure to state a claim. Al-Hamim appealed and continued to represent himself throughout the appeal. His opening brief cited eight “fake cases”, popularly known as “hallucinations.” When the judges hearing Al-Hamim’s cases could not locate the cases as cited, they ordered Al-Hamim to either provide copies of the cited cases or show cause why he shouldn’t be sanctioned for citing fake cases.
To his credit, Al-Hamim admitted that he had used generative AI to help write his opening brief, that the “fake cases” were indeed “hallucinations” created by the generative AI, and that he had failed to check the citations himself before filing his opening brief. Perhaps most important to the outcome of the case, Al-Hamim also apologized and took responsibility for his actions.
After discussing the limitations of generative AI, and the dangers of its use to generate court filings like appellate briefs (these limitations and dangers have been well-documented in cases from other jurisdictions so this post does not delve into that particular issue), the Court of Appeals turned to deciding under what circumstances lawyers and unrepresented litigants may be sanctioned for the misuse of generative AI in Colorado state court proceedings.
The Court began that analysis by anchoring itself to the dictates of Rule 28(a)(7)(B) of the Colorado Appellate Rules. That rule requires that an appellant’s opening brief provide “a clear and concise discussion of the grounds upon which the party relies in seeking a reversal . . . of the judgment . . . of the lower court or tribunal, with citations to the authorities . . . on which the appellant relies.” The Court in Al-Hamim found that the citation of “fake cases” or “hallucinations” violates this rule.
Then, because “no Colorado appellate court has considered the consequences for a self-represented litigant who submits a brief containing hallucinations” the Court turned to decisions from other jurisdictions for guidance. Although other courts have seen fit to sanction pro se litigants for misusing generative AI, the Colorado Court of Appeals declined to do so.
The Court did not issue sanctions for a number of reasons. First, Al-Hamim showed contrition and did not appear to have a history of filing court documents with hallucinated citations. Second, no published Colorado state court decision set out the appropriate sanction for this kind of violation. Third (and perhaps the most teachable moment for lawyers) the Star Hearthstone’s counsel did not separately alert the Court to the hallucinated citations or ask for sanctions.
But, the Court in Al-Hamim closed by warning “lawyers and self-represented parties who appear in this court, that we will not “look kindly on similar infractions in the future.” Lawyers and self-represented litigants in future cases before Colorado state courts may (likely will) now be sanctioned if they use generative AI to generate work product but fail to check its citations.
Of course, lawyers in jurisdictions across the country have been on notice that using generative AI to write briefs is a perilous choice, especially when the lawyer doesn’t check the work product before signing and filing it. But now Colorado lawyers and litigants in particular have been told their use of generative AI is at their own risk.
Al-Hamim also provides a good lesson to parties faced with an opposing party who has cited non-existent authorities: always check your opponent’s citations. One of the factors that led to the Colorado Court of Appeals not sanctioning Al-Hamim is that Star Hearthstone’s counsel did not alert the Court to the hallucinated citations in Al-Hamim’s opening brief or ask for sanctions. While it is hard to know whether the Court would’ve issued a sanction had Star Hearthstone done so, it is good practice to check all authorities your opposing party or counsel cites—and not just to see if they exist, but to check that they actually support the proposition for which they’re cited.
For lawyers, legal research software now has features (which, incidentally, are powered by AI) that will “read” briefs uploaded to them, find the citations in those briefs, and generate a report of whether any of the cited authorities don’t exist and, if they do exist, whether they stand for the proposition for which they’re cited. Al-Hamim thus teaches lawyers that they should regularly be using these powerful time-and-cost saving tools to their advantage. Doing so will not only be to the benefit of them and their clients, but judges, law clerks, and court staff will also likely appreciate being notified of hallucinated citations as they will not need to expend the time, effort, and resources to do so themselves.
This blog post was drafted by Jacob Hollars, a partner in the Spencer Fane Denver office. For more information visit spencerfane.com.
Click here to subscribe to Spencer Fane communications to ensure you receive timely updates like this directly in your inbox.
