Arizona court sanctions lawyer for AI-generated false citations

An Arizona federal court issued extensive sanctions against attorney Maren Bam on August 14, 2025, after finding that her brief contained multiple artificial intelligence-generated citations to non-existent cases. The sanctions include revocation of pro hac vice status, striking the brief, and mandatory notification to state bar authorities.

Judge Alison S. Bachus of the United States District Court for the District of Arizona found that attorney Maren Bam violated Rule 11 of the Federal Rules of Civil Procedure when she submitted a Social Security disability appeal brief containing fabricated legal citations. According to the court's analysis, of 19 case citations in Bam's opening brief, only 5 to 7 cases existed and supported the propositions for which they were cited.

Systematic citation failures uncovered

The court's detailed review revealed three completely fabricated cases attributed to actual Arizona federal judges. Case citations included Hobbs v. Comm'r of Soc. Sec. Admin., Brown v. Colvin, and Wofford v. Berryhill - none of which existed in legal databases. Additional citations misrepresented actual cases or contained quotes that did not appear in the referenced decisions.

According to the court order, Bam's firm had implemented a four-step review process designed to ensure legal integrity and Rule 11 compliance. This process included screening by Bam herself, assignment to contracted attorneys, supervision by on-staff attorneys, and final review by Bam before filing. Despite these procedures, the court found that "failures occurred at every level" of the review process.

The case involved Argelia Esther Mavy's Social Security disability appeal, filed February 28, 2025. Court documents show extensive analysis of similar issues emerging nationwide as AI tools proliferate in legal practice. The court noted that Bam's firm had previously circulated warnings about artificial intelligence use, forwarding a bulletin from the United States District Court for the District of New Mexico that explicitly prohibited AI-generated arguments and citations to non-existent cases.

Extensive sanctions imposed

Judge Bachus imposed six specific sanctions designed to deter similar conduct. The court revoked Bam's pro hac vice status and removed her from the case entirely. Plaintiff's opening brief was stricken from the record. Bam must serve copies of the sanctions order on her client and every judge presiding over cases where she serves as attorney of record.

The sanctions extend beyond the immediate case. Bam must write individual letters to the three Arizona federal judges - Hon. Douglas L. Rayes, Hon. Michael T. Liburdi, and Hon. John Z. Boyle - whose names appeared on the fabricated case citations. The clerk's office will forward the sanctions order to the Washington State Bar Association, where Bam holds license number 46624.

The court's order provides plaintiff Argelia Esther Mavy 45 days to either retain new counsel or proceed as a self-represented party. According to court records, Bam operates a nationwide Social Security disability practice, suggesting the sanctions could affect multiple pending cases across federal courts.

Growing judicial concern over AI hallucinations

The Arizona ruling adds to mounting judicial concern about AI-generated content in legal proceedings. Courts nationwide have documented similar violations, including the landmark Mata v. Avianca case in New York's Southern District, Park v. Kim in the Second Circuit, and recent decisions in California, Texas, and other jurisdictions.

Judge Bachus cited United States v. Hayes from the Eastern District of California, which identified "markings of a hallucinated case created by generative artificial intelligence tools." These markings include case names that appear real but do not exist, identification of actual courts with actual judges' initials, and plausible dates that nonetheless correspond to fictional decisions.

The Second Circuit Court of Appeals established in Park v. Kim that Rule 11 duties require attorneys to "read, and thereby confirm the existence and validity of, the legal authorities on which they rely." The court concluded that submission of briefs relying on non-existent authority reveals failure to determine whether arguments are "legally tenable."

District courts have imposed various sanctions for AI-generated citation violations, including monetary penalties, striking filings, and referral to disciplinary authorities. In Johnson v. Dunn, an Alabama federal court questioned whether monetary sanctions alone provide sufficient deterrence, noting that "if they were effective deterrents, there would not be so many cases to cite."

Enforcement challenges mount

The Arizona case demonstrates enforcement challenges as legal experts warn AI terms of service may prove unenforceable. Legal scholars note that traditional attorney responsibilities cannot be delegated to artificial intelligence systems or support staff, regardless of technological sophistication.

According to the court's analysis, Bam's response emphasized her reliance on contracted attorneys and supervisory staff to prepare and review the brief. However, Rule 11 establishes that "the signing attorney cannot leave it to some trusted subordinate, or to one of his partners, to satisfy himself that the filed paper is factually and legally responsible."

The ruling reflects broader concerns about AI compliance in professional settings. Federal agencies have begun enforcement actions against companies exploiting AI capabilities for deceptive practices, while courts grapple with maintaining evidentiary standards amid increasing AI adoption.

Professional responsibility rules require attorneys to maintain competence and exercise independent professional judgment. These obligations persist regardless of technological tools employed in legal practice. The Arizona court emphasized that "a lawyer who wishes to use AI ethically must ensure that the legal propositions and authority generated are trustworthy."

Industry response and future implications

The sanctions order notes that "pitfalls of utilizing AI are now well-known in this profession," citing growing awareness of generative AI limitations in legal contexts. Multiple federal courts have implemented AI disclosure requirements, while state bar associations develop guidance for ethical AI use.

Recent regulatory developments highlight challenges courts face as AI systems demonstrate increasing autonomy. Legal frameworks struggle to assign responsibility when AI tools produce unreliable outputs that attorneys fail to verify independently.

The marketing community faces similar accessibility concerns with AI implementations, as technology companies deploy AI systems without adequate verification mechanisms. Professional service providers across industries encounter pressure to adopt AI tools while maintaining established quality standards.

The Arizona court's comprehensive sanctions signal judicial determination to preserve legal system integrity despite technological advancement. The order serves as a warning that professional licensing privileges and client representation rights depend on maintaining traditional verification standards regardless of AI tool availability.

Timeline

• February 28, 2025: Argelia Esther Mavy files Social Security disability appeal complaint through attorney Maren Bam
• May 27, 2025: Bam files opening brief containing multiple fabricated citations
• June 12, 2025: Social Security Administration files response brief
• July 22, 2025: Court issues Order to Show Cause regarding citation deficiencies
• July 31, 2025: Bam responds to show cause order, taking responsibility but emphasizing reliance on contractors
• August 14, 2025: Judge Bachus issues sanctions order finding Rule 11 violations
• January 2025: Minnesota court excludes Stanford professor's testimony over AI-generated citations
• December 2024: Illinois Supreme Court sets lenient AI disclosure requirements
• September 2024: FTC launches Operation AI Comply targeting deceptive AI practices

PPC Land explains

Rule 11 Sanctions: Federal procedural rule requiring attorneys to certify that legal contentions are warranted by existing law after conducting reasonable inquiry. Violations can result in monetary penalties, striking of filings, attorney disqualification, or referral to disciplinary authorities. The rule establishes objective standards for attorney conduct, meaning violations do not require subjective bad faith. Courts apply Rule 11 with particular stringency when imposing sua sponte sanctions, requiring conduct "akin to contempt of court."

Pro Hac Vice Status: Legal privilege allowing attorneys licensed in one jurisdiction to appear in courts of another jurisdiction for specific cases. The term means "for this occasion only" in Latin. Admission pro hac vice is discretionary and can be revoked for failure to comply with federal or local rules. The Arizona case demonstrates that pro hac vice revocation serves as both punishment and deterrent, particularly impacting attorneys with nationwide practices who routinely seek such admissions.

AI Hallucinations: Artificial intelligence-generated content that appears authoritative but contains fabricated information, particularly fake legal citations that reference non-existent cases with realistic-sounding names, actual court identifications, and plausible dates. Legal AI hallucinations have become a significant concern as generative AI tools proliferate in legal practice, leading courts to develop specific identification criteria and enforcement responses.

Citation Fabrication: Creation of false legal references that misrepresent existing law or cite non-existent authorities. In the Arizona case, fabricated citations included three completely fictitious cases attributed to real federal judges, plus multiple existing cases misquoted or cited for unsupported propositions. Citation fabrication violates fundamental attorney duties to verify legal authorities and constitutes a false statement of law to courts.

Brief Striking: Court remedy that removes deficient legal filings from the official record, effectively nullifying the document's legal effect. When courts strike briefs containing AI-generated citations, they often find that removing fabricated authorities leaves insufficient legitimate support for legal arguments. Brief striking serves both punitive and practical functions, eliminating unreliable legal submissions while requiring parties to start over with verified authorities.

Attorney Sanctions: Disciplinary measures imposed by courts for professional misconduct, ranging from monetary fines and public reprimands to case removal and bar referrals. In AI hallucination cases, sanctions typically include multiple components designed to address both individual accountability and systemic deterrence. The Arizona ruling's six-part sanctions package reflects courts' determination to prevent proliferation of AI-generated citation violations.

Legal Authority Verification: Professional obligation requiring attorneys to confirm the existence, accuracy, and relevance of cited legal sources before filing court documents. This duty cannot be delegated to support staff, contractors, or artificial intelligence systems. Verification requirements have gained prominence as AI tools generate increasingly sophisticated but unreliable legal references that require independent confirmation through traditional legal research methods.

Federal Rules of Civil Procedure: Comprehensive framework governing civil litigation in United States district courts, including Rule 11's attorney certification requirements. These rules establish uniform standards for legal practice across federal jurisdictions, with local courts authorized to impose additional requirements. The rules' application to AI-generated content reflects courts' adaptation of existing frameworks to address technological challenges.

Bar Association Referral: Process by which courts notify attorney licensing authorities of professional misconduct, potentially triggering disciplinary proceedings separate from immediate case sanctions. Bar referrals serve broader regulatory functions beyond individual case resolution, enabling state licensing bodies to monitor attorney conduct patterns and implement professional education or disciplinary measures as appropriate.

Social Security Disability Appeals: Specialized federal court proceedings challenging administrative denials of disability benefits, governed by specific evidentiary and procedural requirements. These cases often involve extensive medical evidence and regulatory interpretation, making accurate legal citation particularly important for establishing appellate standards and precedential authority. The Arizona case's Social Security context highlights AI reliability concerns in technical legal practice areas.

Summary

Who: Attorney Maren Bam from Washington State (Bar No. 46624) operating a nationwide Social Security disability practice, sanctioned by United States Magistrate Judge Alison S. Bachus in the District of Arizona.

What: Comprehensive Rule 11 sanctions including pro hac vice revocation, brief striking, mandatory client notification, bar association referral, and required letters to judges whose names appeared on fabricated citations.

When: Sanctions imposed August 14, 2025, following opening brief filed May 27, 2025, and show cause proceedings initiated July 22, 2025.

Where: United States District Court for the District of Arizona in case No. CV-25-00689-PHX-KML, with implications for Bam's nationwide practice and pending cases across federal courts.

Why: Court found majority of 19 legal citations were fabricated by artificial intelligence, with only 5-7 cases existing and supporting stated propositions, violating attorney duties to conduct reasonable inquiry into legal authorities before filing.