Skip to content
AID-2025-00042025-2026high

MyPillow Attorney Filings with Fabricated Citations and a Repeat Miscitation

Attorneys filed a brief with AI-generated citations to nonexistent cases and were sanctioned; months later the lead attorney was sanctioned again for misciting a real case to the wrong court and mischaracterizing its holding.

FabricationNamed approval gate

What happened

In 2025, attorneys for Mike Lindell filed a response brief in a Denver federal case (Coomer v. Lindell) that contained nearly 30 defective citations, including citations to nonexistent cases and misquotations of real cases. Christopher Kachouroff admitted he had used an AI tool while drafting that brief. Judge Nina Wang sanctioned Kachouroff and his law firm $3,000 and co-counsel Jennifer DeMaster $3,000 in July 2025. In May 2026, the court sanctioned Kachouroff a further $5,000 for a separate error in a later post-trial brief: he cited a real case, Capital Solutions, LLC v. Konica Minolta Business Solutions USA, Inc., but attributed that District of Kansas decision to the Tenth Circuit and mischaracterized its holding. Kachouroff denied using generative AI for the second brief, and the court made no finding that AI caused it, ruling only that the citation was defective and should have been caught by a reasonable review regardless of whether AI was used. The court declined to refer Kachouroff to the Virginia bar for discipline.

What the agent did

A human attorney filed the briefs. The citations were not independently verified before filing. The first brief contained nonexistent cases (AI use in drafting was admitted); the second contained a real case attributed to the wrong court and mischaracterized (AI involvement was disputed and not confirmed by the court).

The irreversible effect

Defective citations entered the court record: nonexistent cases in the first brief and a real case miscited and mischaracterized in the second. Monetary sanctions were imposed ($3,000 each on Kachouroff and DeMaster in the first instance, $5,000 on Kachouroff in the second), with reputational harm. The judge declined to refer the attorneys for bar discipline.

Root cause

No independent verification of the cited cases occurred before filing. A reasonable inquiry, or review by a second person, would have confirmed whether each cited case existed, came from the court it was attributed to, and stood for the stated proposition. The first sanction did not lead to a citation-verification step, so a similar error recurred in a later filing.

How a maker-checker control would have refused it

approval_gate: Court filings route through a maker-checker approval gate that requires an independent reviewer to verify every cited case (that it exists, comes from the court it is attributed to, and stands for the stated proposition) against the source before the brief can be filed. Because a human, not an agent, filed these briefs, the relevant control is independent citation verification by a second person, not an automated block on the filing action. Such review would have caught the nonexistent cases in the first brief and the miscited, mischaracterized case in the second.

Runnable reproduction

This incident ships as a runnable scenario in the open-source repository. Point the enforcement engine at the policy and watch the action get refused, with the refusal written to a signed audit record.

examples/mypillow-ai-brief-fake-citations-repeat

View the reproduction on GitHub →

Accuracy and corrections

This entry describes a publicly reported incident and is compiled from the primary sources listed above. Where an account is a legal allegation rather than an established finding, the entry labels it as such. Summaries can still contain errors. If you can document a correction, email hello@makerchecker.ai and we will review and correct it, with the change noted, within 14 days.

See it for yourself

Reading is one thing. Watch it block an agent.

One command starts the demo: an agent stopped from signing off its own work, and the signed evidence file an inspector can check for themselves.

Designed against the rules your auditors already enforce.