What Happens If You Cite a Hallucinated Case in a Filing

Written by the Citation Safe Research Desk · Reviewed by Andy Gaber, Founder — July 16, 2026

The consequences of filing a hallucinated citation vary by jurisdiction and by how the court discovers the problem, but the Damien Charlotin AI Hallucination Cases database, tracking well over 1,700 documented cases, gives a realistic picture of the range, not a hypothetical one.

The immediate procedural consequence

Most commonly, the offending motion or brief is struck in whole or in part, or the argument relying on the fabricated citation is disregarded. In some documented cases the underlying claim or defense is dismissed outright when the fabricated citations were central to it. Courts have also required refiling with a certification that every citation has been independently checked.

Monetary sanctions

Monetary sanctions in the documented cases range from a few hundred dollars to tens of thousands, occasionally covering the opposing party's attorney's fees incurred responding to the fabricated citations. In Mata v. Avianca, the court imposed a monetary sanction on the attorneys involved and required notice to the presiding judges in every case the fabricated opinions had purportedly been drawn from. In Kruse v. Karlen (Mo. Ct. App. 2025), a pro se litigant was sanctioned $10,000 after submitting briefing with AI-fabricated citations — a reminder that courts do not automatically extend leniency to self-represented parties on this issue. Courts have also ordered attorneys to personally bear costs rather than passing them to the client, on the theory that the client did not cause the failure to verify.

Professional discipline

Bar referrals appear frequently in the documented cases, ranging from a referral for review to reported disciplinary action. In Park v. Kim, 91 F.4th 610 (2d Cir. 2024), the Second Circuit referred the attorney's conduct for further review after identifying a citation to a case that did not exist. Some orders require the attorney to notify the client, notify co-counsel, or notify the court of the same failure in other pending matters.

Reputational and case consequences

Once a sanction order is public (and many are, since courts frequently publish these decisions specifically as a deterrent), it is searchable by opposing counsel in future matters, by potential clients, and by legal press covering the AI-hallucination story. Several of the documented cases, including Wadsworth v. Walmart Inc. (D. Wyo. 2025) and Coomer v. Lindell (D. Colo.), were covered by national legal press specifically because of the AI angle, extending the reputational exposure well beyond the courtroom.

Expert perspective

Charlotin's own commentary on the database he maintains notes that the severity of sanctions correlates less with the number of fabricated citations and more with the attorney's response once the problem was flagged: prompt correction and candor with the court consistently draw more lenient treatment than defending the citations after they are challenged.

What this means practically

None of these consequences are speculative or theoretical — they are drawn from a public, growing record. The fix is not "stop using AI to draft" (most of the profession is not going to do that), it is verifying every citation against a primary source before filing, which is the one step nearly every documented sanction case shows was skipped.

A quick self-check before you file

Has every citation been checked for existence against a primary source?
Has every quotation been checked word-for-word?
Have you confirmed the case is still good law on the point cited?
If you found an error, have you corrected it and disclosed it, rather than hoping it goes unnoticed?

A common question

Does it matter whether the fabrication was 'my fault' versus the AI tool's?

From a Rule 11 standpoint, generally not much — the signing attorney bears the certification duty regardless of which tool assisted in drafting. Courts have been fairly consistent that using an AI tool does not shift or dilute that duty. See our explainer on Rule 11 and AI-generated filings for the full framework.

How consequences compound with repeated incidents

First-time incidents, especially those self-corrected promptly, have generally drawn more lenient treatment across the documented cases than repeated incidents by the same attorney or firm. Courts and bar disciplinary bodies have shown less patience for a second documented AI-hallucination incident, treating it as evidence the first incident did not produce meaningful process changes. This pattern is worth internalizing: the consequences of a first mistake are not just about that single filing, they are also about establishing whether your practice actually learned from it.

Insurance and coverage implications

A sanctions order or malpractice claim arising from a hallucinated citation can trigger reporting obligations under your malpractice policy, and depending on your policy's specific terms, may affect future coverage terms or premiums. See our detailed piece on what malpractice insurers are saying about AI drafting risk for how carriers are treating this exposure today.

Final takeaway

Every consequence described in this piece traces back to the same single missed step: independent verification against a primary source before filing. That is both the bad news (these consequences are all avoidable failures, not bad luck) and the good news (the fix costs a few minutes per citation using free tools).

If you take away one thing from this piece, let it be that none of these outcomes are randomly distributed bad luck; every documented case traces back to a specific, identifiable point where independent verification could have caught the problem before it reached a judge.

The five minutes it takes to verify a brief before filing is, without exception, cheaper than any outcome described above.

Make it a standing part of your filing process starting today, not after your own name shows up in a public tracker.

A documented, consistent habit is the only defense the public record has shown to actually work, across every practice area and every AI tool reviewed.