SlopJudges
Tracking AI slop in real U.S. Supreme Court, Courts of Appeals, and federal District Court opinions. Every opinion is pulled from the public CourtListener record and scored by our own blend of an LLM judge and deterministic heuristics — no third-party detector.
- Opinions analyzed
- 3,128
- Slop opinions
- 98
- Courts tracked
- 26
- Last updated
- Nov 15, 2026, 2:42 PM
Cumulative AI-slop opinions over time
Supreme Court vs. the regional Courts of Appeals vs. the specialized D.C. & Federal Circuits vs. the federal District Courts.
Justices by AI-slop opinions
Supreme Court justices ranked by opinions flagged as slop; ties break on average AI score.
- 1
- 2
- 3
- 4
- 5
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15
- 16
- 17
- 18
Courts by AI-slop opinions
All tracked courts ranked by opinions flagged as slop; ties break on average AI score.
- 1
- 2
- 3
- 4
- 5AI 472nd Cir.4 avg AI · 171 opinions↳ Petersen Energ�a; Eton Park v. Argentie Argentine Republic, YPF S.A.6
- 6
- 7
- 8
- 9
- 10
- 11
- 12
- 13
- 14
- 15AI 19Fed. Cir.4 avg AI · 69 opinions↳ Teva Pharmaceuticals International Gmbh v. Eli Lilly and Company1
- 16
- 17
- 18
- 19
- 20
- 21
- 22
- 23
- 24
- 25
- 26AI 0N.D. Ill.0 avg AI · 0 opinions0
Most AI-like opinions
The highest-scoring opinions in the dataset. On the real federal record almost none cross the flag line (18); these are simply the current leaders, shown with the signals that moved their score.
- United Financial Casualty Co. v. Greg BallAI 50Court of Appeals for the Fourth Circuit · Mar 30, 2023
The text consists solely of repeated document metadata with no substantive legal content, indicating it is not a genuine judicial opinion.
repeated document metadatalack of substantive contentLLM 100Heuristic 0Confidence 20% - Petersen Energ�a; Eton Park v. Argentie Argentine Republic, YPF S.A.AI 47Court of Appeals for the Second Circuit · Mar 27, 2026
The text contains malformed citations and potentially fabricated case names, indicating a high likelihood of AI assistance.
malformed reporter citationsfabricated case namesLLM 85Heuristic 12Confidence 18% - American Academy of Pediatrics v. James UthmeierAI 37Court of Appeals for the Seventh Circuit · Sep 14, 2026 · Scudderconcurs and dissents
Although much of the prose resembles authentic appellate writing, the co-occurrence of serious citation anomalies, likely fabricated future citations, and some polished essay-style structuring makes substantial AI generation likely.
citation problems and apparent hallucinationswrong Dombrowski citation volume repeatedimpossible cert-denied date preceding appellate decisionsome essay-like framing and tidy Roman-numeral structureLLM-favored phrases such as 'At its core' and broad thematic framingLLM 78Heuristic 10Confidence 86% - United States v. SeniorAI 35District Court, District of Columbia · Oct 6, 2026 · Judge Colleen Kollar-Kotelly
Heuristic-only score (LLM judge unavailable).
Unusually uniform sentence rhythmSparse citation density for opinion lengthLLM 0Heuristic 35Confidence 30% - In re: E. Palestine Train DerailmentAI 32Court of Appeals for the Sixth Circuit · Nov 10, 2026 · Amul R. Thapar; Chad A. Readler; Whitney D. Hermandorfer
The excerpt reads like a polished appellate opinion with concrete record citations, varied punchy and clause-heavy sentences, and no obvious LLM boilerplate or generic essay structure, though the future/unverifiable citations slightly limit confidence.
Unusually uniform sentence rhythmSparse citation density for opinion lengthprecise and conventionally formatted citationsvaried sentence lengths and judicial rhythmno chatbot boilerplate or listicle scaffoldingLLM 10Heuristic 37Confidence 82% - United States v. Alexzandra BlancoAI 31Court of Appeals for the Eighth Circuit · Aug 10, 2026
The text otherwise reads like a conventional Eighth Circuit opinion with concrete facts, standard formatting, precise statutory/rule citations, and no chatbot-style framing or buzzword clusters, but the malformed Campbell citation is a notable isolated AI-like warning sign.
Unusually uniform sentence rhythmisolated impossible reporter/year citation: 6 F.4th with 2014 datesome repetitive appellate-disposition phrasingLLM 22Heuristic 33Confidence 78% - Cassandra Sykes v. Experian Information Solutions, Inc.AI 31Court of Appeals for the Seventh Circuit · Jul 30, 2026 · Jackson-Akiwumi
The excerpt reads like a conventional appellate opinion with precise and plausible citations, normal judicial structure, varied sentence rhythm, and no chatbot boilerplate, buzzword clustering, listicle scaffolding, or citation anomalies apparent from the text.
Unusually uniform sentence rhythmChatbot-style holding recapLLM 5Heuristic 38Confidence 90% - L.M. Caheb v. MullinAI 30District Court, District of Columbia · Nov 10, 2026 · Chief Judge James E. Boasberg
The excerpt is too short and truncated to assess meaningfully, but it shows ordinary court-caption and introductory-opinion language with no obvious generative-AI tells.
Unusually uniform sentence rhythmLLM 5Heuristic 35Confidence 2% - Patel v. Atlantic Monthly Group LLCAI 29District Court, District of Columbia · Nov 10, 2026 · Judge Emmet G. Sullivan
The excerpt is only a caption and opening sentence with standard judicial formatting and no discernible generative-AI tells, making any assessment highly uncertain.
Unusually uniform sentence rhythmLLM 5Heuristic 34Confidence 2% - Ward v. MayorkasAI 29District Court, District of Columbia · Sep 11, 2026 · Judge Reggie B. Walton
Although the prose and structure otherwise resemble a genuine D.D.C. memorandum opinion, the assertion that Markwayne Mullin is the current Secretary of Homeland Security and the related substitution history are blatant factual/citation-style anomalies that strongly suggest synthetic or substantially AI-assisted drafting.
fabricated or impossible official-capacity defendant statementsuspicious party substitution footnotepossibly hallucinated future/implausible citationotherwise polished opinion-like formatting with precise but questionable citationsLLM 72Heuristic 0Confidence 82% - Rapaport v. NivodaAI 29Court of Appeals for the Second Circuit · Sep 4, 2026
Heuristic-only score (LLM judge unavailable).
Unusually uniform sentence rhythmLLM 0Heuristic 29Confidence 30% - Ronda L. Davis v. DCAI 29Court of Appeals for the D.C. Circuit · Sep 4, 2026
Heuristic-only score (LLM judge unavailable).
Unusually uniform sentence rhythmLLM 0Heuristic 29Confidence 30% - Maria Reynoso-Salgado v. Todd W. BlancheAI 27Court of Appeals for the Seventh Circuit · Sep 1, 2026 · Jackson-Akiwumi
The prose largely resembles a genuine appellate opinion, but multiple citation anomalies—especially the impossible-looking pin cites—are strong machine-generation tells that raise the likelihood of substantial AI assistance.
malformed pinpoint citation “1191–13”internal citation inconsistency for State v. Evans, 492 N.W.2d 141 later cited at 481several post-2024 citations and party names that appear potentially fabricated or difficult to verify from form alonevery tidy categorical-approach structureLLM 58Heuristic 7Confidence 78% - Liu v. NoemAI 27District Court, District of Columbia · Jul 29, 2026 · Judge Jia M. Cobb
The prose largely reads like a conventional judicial opinion, but the highly implausible official-capacity defendant substitution and possibly fabricated future case citations are strong hallucination-style signals that substantially raise the likelihood of AI assistance.
implausible party substitution naming Markwayne Mullin as Secretary of Homeland Securitypossible fabricated or anachronistic supporting cases with 2026 WL citationsotherwise precise but potentially hallucinated citation scaffoldingno chatbot boilerplate or obvious buzzword clustersLLM 68Heuristic 1Confidence 78% - Perales v. Office of the ClerkAI 27District Court, District of Columbia · Jul 23, 2026 · Judge Jia M. Cobb
The excerpt is a very short, standard dismissal order with no chatbot phrasing, buzzword clusters, citation issues, or structural AI tells, though its brevity limits confidence.
Unusually uniform sentence rhythmLLM 2Heuristic 32Confidence 7% - Truesdale v. District of Columbia GovernmentAI 26District Court, District of Columbia · Oct 23, 2026 · Judge Dabney L. Friedrich
The order is concise, case-specific, and conventionally judicial with precise docket references, but a notable impossible reporter citation and a stray drafting error create some suspicion without amounting to strong LLM evidence.
Unusually uniform sentence rhythmSparse citation density for opinion lengthone impossible/malformed citation: F.3d reporter with 1964 dateminor proofreading error: 'under the of professional conduct'otherwise concrete docket references and case-specific reasoningLLM 22Heuristic 27Confidence 78% - Public Citizen v. United States Department of JusticeAI 26District Court, District of Columbia · Oct 22, 2026 · Judge Trevor N. McFadden
The text reads like a conventional short federal FOIA order with precise docket references, plausible and properly formatted citations, normal judicial phrasing, and no chatbot boilerplate, buzzword clustering, listicle structure, or obvious citation hallucinations.
Unusually uniform sentence rhythmChatbot-style holding recapLLM 8Heuristic 30Confidence 86% - United States v. Davone WalkerAI 26Court of Appeals for the Third Circuit · Sep 15, 2026
This brief procedural order uses standard appellate formatting and precise docket details with no chatbot phrasing, buzzword filler, mechanical essay structure, or citation anomalies.
Unusually uniform sentence rhythmLLM 2Heuristic 32Confidence 20% - United States v. Calvin CarterAI 26Court of Appeals for the Eighth Circuit · Jul 7, 2026
The excerpt reads like a conventional concise appellate per curiam opinion with precise Guidelines and case citations, fact-specific reasoning, and no chatbot boilerplate, buzzword clusters, vague legal generalities, or malformed citation patterns suggestive of LLM drafting.
Unusually uniform sentence rhythmChatbot-style holding recapLLM 5Heuristic 32Confidence 93% - United States v. Abdul KareemAI 25Court of Appeals for the Ninth Circuit · Nov 6, 2026
Although the format mimics a Ninth Circuit opinion, the text contains several independent machine-generation tells, including a conspicuously wrong rule citation, a likely hallucinated terrorism attribution, and mechanically repetitive opinion-style prose.
factually suspect attribution of Anwar al-Awlaki as an ISIS leaderapparently wrong appellate rule citation for a criminal appeal: Fed. R. App. P. 4(a)(1)(A)repetitive mechanical factual structure with A-F subsections repeatedly beginning with government-introduced evidencegeneric essay-like opening and tidy framingunusual vague phrasing such as ISIS-related or ISIS-adjacent materialsLLM 62Heuristic 1Confidence 78% - American Academy of Pediatrics v. James UthmeierAI 25Court of Appeals for the Seventh Circuit · Sep 14, 2026 · Kolarconcurs
The prose mostly resembles a genuine appellate opinion with varied rhythm and many precise citations, but the internally implausible and erroneous citations are strong enough hallucination-like signals to put it in a moderate-risk range.
one clear erroneous citation to Dombrowski as 389 U.S. 479 instead of 380 U.S. 479implausible certiorari citation dated before the lower-court decisionsome essay-like tidy structuring with numbered sections and brief concluding paragraphsoccasional LLM-favored phrasing such as “safety valve” and “underscoring”LLM 48Heuristic 10Confidence 74% - Annette Kingsolver v. U.S. Attorney GeneralAI 25Court of Appeals for the Eleventh Circuit · Aug 20, 2026
The excerpt reads like a conventional appellate opinion with precise, plausible citations, case-specific factual detail, varied judicial sentence structure, and no chatbot framing, buzzword clusters, malformed citations, or essay-like AI scaffolding.
Unusually uniform sentence rhythmLLM 8Heuristic 29Confidence 86% - Dunn v. Santa Fe Natural Tobacco CompanyAI 24Court of Appeals for the Tenth Circuit · Nov 12, 2026
The text reads like a conventional appellate order and dissent with precise legal issues, plausible citations, varied sentence structure, and no chatbot boilerplate, buzzword clustering, malformed citations, or generic essay-style filler.
Unusually uniform sentence rhythmLLM 6Heuristic 28Confidence 90% - Students Engaged in Advancing Texas v. Ken PaxtonAI 24Court of Appeals for the Fifth Circuit · Oct 28, 2026
This is a short, formulaic en banc rehearing order with standard court formatting, precise docket references, and no chatbot phrasing, buzzword clusters, mechanical essay structure, or citation irregularities.
Unusually uniform sentence rhythmHeavy em-dash usageLLM 2Heuristic 29Confidence 48% - Perez v. v Wax Studio, LLCAI 24District Court, District of Columbia · Sep 24, 2026 · Judge Rudolph Contreras
The excerpt is largely plausible law-clerk style, but the cluster of suspect citations, citation-format anomalies, mechanical structure, and small procedural/formatting inconsistencies suggests a meaningful likelihood of substantial generative-AI assistance.
multiple suspicious or possibly fabricated 2026 WL citationsmalformed citation formatting such as "503 Supp.2d"mechanical judicial template and tidy sectioninginconsistent numbering and capitalization conventionsunusual phrasing such as "For the forthcoming reasons"LLM 62Heuristic 0Confidence 74%