Sanction teardown · E.D. Oklahoma, USA · 2025-10-22

Mattox v. Product Innovation Research

What happened

In E.D. Oklahoma, USA, a filing relied on ChatGPT to help draft legal argument. The court identified the following problems with the citations in that filing:

• Fabricated (Case Law)
  Court found no case under this citation; similar Tenth Circuit decision exists but does not contain the quoted law.
• Fabricated (Case Law)
  Citation combined/ conflated unrelated authorities; court found no case at that citation and similar decisions do not support the quoted law.
• Fabricated (Case Law)
  Court found no case under the cited name and citation; citation appears fabricated/mismatched.
• Fabricated (Case Law)
  No Tenth Circuit decision exists at that citation; the docketed citation corresponds to an unrelated out-of-circuit case.
• Fabricated (Case Law)
  Court found no case under that citation; similar named district decision exists but does not support plaintiffs' asserted rule.
• Fabricated (Case Law)
  Court located no case under this citation; authority appears fabricated.
• Misrepresented (Case Law)
  Plaintiffs cited the wrong year/citation for the decision; court identified correct authority.
• False Quotes (Case Law)
  Court determined the specific language plaintiffs attributed to the case does not appear in the opinion cited.
• False Quotes (Case Law)
  Court found the law plaintiffs claimed derived from this Supreme Court opinion was not present in the opinion as quoted.
• Misrepresented (Case Law)
  Court concluded the decision does not hold the proposition plaintiffs attributed to it.
• Fabricated (Case Law)
  Court found no decision at this federal reporter/citation; cited authority is incorrect or nonexistent.
• Misrepresented (Case Law)
  Plaintiffs cited an incorrect citation; the court identified the correct citation and that the cited version was erroneous.

Which AI tool

ChatGPT. Note: Charlotin's public database records tool attribution only where a court order, brief, or reporting on the matter states it explicitly; "unidentified" or "implied" means the record indicates AI use but does not name a specific product — we do not guess.

Outcome

Pleadings struck; public reprimands; monetary sanctions; remedial filing and certification requirements (monetary penalty: 28495 USD.)

Additional detail

The Court found 28 false or misleading citations across 11 pleadings (14 fabricated, 14 erroneous/misquoted). Mr. Howie admitted use of ChatGPT and failure to verify citations. The Court applied Rule 11(b) and its AI framework (verification, candor/correction, accountability) and imposed sanctions and restitution. Fines of 3,000, 2,000, and 1,000 USD on individual attorneys, plus opposing party's costs and fees,

How Citation Safe would have caught this

Citation Safe runs three deterministic layers before a brief is filed: (1) does the citation exist against CourtListener's database of published opinions, (2) if quoted, does that exact language appear in the source, (3) does the cited case actually support the proposition it is cited for. Fabricated case citations fail Layer 1. Fabricated or misattributed quotations fail Layer 2 even when the underlying case is real. Misrepresented holdings — a real case cited for a proposition it does not support — are the target of Layer 3. None of these checks involve asking another language model whether the citation looks right; they are lookups and text-matches against the actual source, which is why a hallucinated citation has to survive a direct lookup against the authoritative source — not another model's opinion — to earn a VERIFIED stamp; our measured false-verify rate is published live at /quality.

Check a brief before you file it → · See our live false-verify rate

Source: https://www.damiencharlotin.com/documents/904/Mattox_v._Product_Innovations_Research_USA_22_October_2025.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).