Sanction teardown · M.D. Florida, USA · 2025-05-15

USA v. Burke

What happened

In M.D. Florida, USA, a filing relied on Westlaw's AI tools, GPT4.5 Deep Research (Pro) to help draft legal argument. The court identified the following problems with the citations in that filing:

• False Quotes (Case Law)
  Attributed a quote to Ruiz that does not exist and used it for an unsupported proposition.
• Misrepresented (Case Law)
  Claimed Ruiz supports the rule that statutes cannot require defendants to prove exculpatory facts, which the case does not support.
• False Quotes (Doctrinal Work)
  Attributed to Senator McClellan a statement on telephone privacy not supported by the cited Congressional Record entry.
• False Quotes (Case Law)
  Attributed to Doe v. Miami-Dade County a quote about constitutional avoidance being strongest when protecting fundamental rights, which is not stated there.
• False Quotes (Case Law)
  Falsely quoted McQueen as stating the government must prove each element beyond a reasonable doubt.
• Misrepresented (Case Law)
  Claimed McQueen supports that indictment ambiguities or omissions violating preparation or double-jeopardy interests violate the Sixth and Fifth Amendments, which McQueen does not support.
• False Quotes (Case Law)
  Misattributed a district court’s quote to the Ninth Circuit in Konop regarding the 'readily accessible' exception.
• False Quotes (Case Law)
  Quoted Innovatio but omitted half of the sentence actually appearing in that case.
• False Quotes (Case Law)
  Falsely quoted Gray as saying it is impermissible to require a defendant to prove a negative essential to innocence.
• Misrepresented (Case Law)
  Cited Green for the proposition that the government must disprove statutory exceptions, which the case does not support.
• Misrepresented (Doctrinal Work)
  Relied on a House Judiciary Report that concerns the Flammable Fabrics Act Amendments, not the relevant legislation.
• False Quotes (Case Law)
  Falsely quoted Tsurkan as stating the Wiretap Act's primary objective is protecting telephone conversations from interception.
• Misrepresented (Case Law)
  Miscited Tsurkan; provided a reporter citation that corresponds to a different case and not the quoted material.
• False Quotes (Case Law)
  Attributed to Hasson a constitutional avoidance quote not stated there.
• False Quotes (Case Law)
  Falsely quoted Veal as forbidding courts from ignoring statutory definitions in favor of dictionary meanings.
• False Quotes (Case Law)
  Falsely quoted Wisconsin Right to Life with a definition of as-applied challenges not found in the opinion.
• False Quotes (Doctrinal Work)
  Attributed to a 1968 Senate Report a definition of 'wire communication' not present in the report.
• False Quotes (Doctrinal Work)
  Attributed to a 1986 Senate Report a statement about public broadcasts not found there.
• False Quotes (Case Law)
  Attributed to Harrell a quoted holding about as-applied unconstitutionality not actually quoted in the case.
• False Quotes (Case Law)
  Attributed to Harrell a long proposition about privacy versus public speech and press exemptions not actually stated there.
• False Quotes (Case Law)
  Attributed to Bobo a quoted holding on constructive amendments and vague charges not contained in the opinion.

Which AI tool

Westlaw's AI tools, GPT4.5 Deep Research (Pro). 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

Motion dismissed, and plaintiff ordered to refile it without fake citations.

Additional detail

Counsel later explained how the motion came to be: see here.

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/438/USA_v._Burke_M.D._Florida_USA_May_15_2025.pdf, via Damien Charlotin's public AI Hallucination Cases Database (CC0).