AI

Fake AI Citations Strike Mortgage Trustee's Foreclosure Brief

Deutsche Bank left without appellate filing after outside counsel used Google's AI tool and failed to verify four nonexistent cases.

Omega Editorial· September 10, 2026· 4 min read

Appellate Court Strikes Filing Over Fabricated Cases

A mortgage trustee defending a foreclosure judgment was left without an appellate brief after its outside attorneys submitted citations to four court decisions that do not exist. The District of Columbia Court of Appeals struck the entire filing submitted on behalf of Deutsche Bank National Trust Company, acting as trustee for a residential mortgage-backed securities trust.

The three-judge panel issued a published order on September 3 that did not overturn the underlying foreclosure judgment or accuse Deutsche Bank of wrongdoing. Instead, it exposed how inadequate AI controls at an outside law firm can create litigation risk for the mortgage companies, trusts, and investors they represent in court.

"What began as a routine appeal has become a cautionary tale about the misuse of artificial intelligence," the panel wrote.

How the Error Occurred

The case stems from a judicial foreclosure action against homeowner Barry Douglas, who is representing himself on appeal. While reviewing the parties' briefs, the appellate court discovered it could not locate or verify multiple authorities cited in Deutsche Bank's filing.

Attorney Loishirl W. Hall, who filed the brief, confirmed that four cited cases did not exist. According to the court order, Hall acknowledged using Google's generative AI search tool to locate case authority without verifying the citations before filing. Hall told the court it was her first appellate brief and has since left McCabe, Weisberg & Conway LLC, Deutsche Bank's outside counsel.

Two other attorneys were listed on the brief. The firm stated it was unaware of Hall's actions when the filing was submitted and that its policies prohibit employees from using AI to draft legal documents. The firm also requires citations to be verified regardless of how they were obtained.

The court rejected the firm's request to accept an amended brief with the false citations removed and struck the entire filing. The panel noted that every attorney who signed the brief bore some responsibility for its contents, though Senior Judge Stephen Glickman clarified this does not require each attorney to personally check every citation if reasonable verification procedures exist.

The matter was referred to the Office of Disciplinary Counsel for potential investigation, though a referral does not constitute a finding of misconduct. Glickman characterized the conduct as negligent or grossly negligent based on the available record.

Why it matters

Mortgage servicers manage foreclosure litigation for investors and securitized trusts through outside law firms. A breakdown at one of those firms can delay proceedings, increase legal costs, or undermine a trust's ability to defend a favorable ruling. This case demonstrates that AI risk extends beyond a company's own employees and technology vendors to every outside professional whose work affects lien enforcement and litigation outcomes.

Controls for Third-Party AI Use

Mortgage businesses evaluating AI risk should consider controls that address outside counsel and service providers. Relevant measures may include requiring disclosure of generative AI use on mortgage-related matters, independent verification of AI-assisted research, supervisory review before court filings, prompt reporting of AI-related errors, and clear assignment of responsibility for resulting costs and delays.

The law firm in this case already had an AI policy in place. The filing nevertheless reached an appellate court with four fabricated cases. A written policy is not an effective control unless someone verifies compliance.

A similar problem emerged in January when a New York appellate court imposed $10,000 in combined sanctions after a borrower and his attorney submitted at least 23 fabricated cases across five filings. Together, these cases show that AI-generated errors can disrupt foreclosure litigation regardless of which party introduces them.

The details were first reported by National Mortgage Professional.

#mortgage foreclosure#ai litigation risk#legal ai errors#mortgage servicing#third-party oversight#generative ai

This is an original analysis by the Omega editorial team. Source reporting: AI Watch.

Want systems like this working for your business?

Book a Call

More in AI

AI· 3 min read

Positron AI raises $875M for memory-first inference chips

The Nevada startup's valuation jumped to $5 billion as it prepares next-generation hardware designed to handle trillion-parameter models.

Via AI Watch · Sep 10, 2026
AI· 3 min read

NASA Releases Open-Source AI Model Trained on Lunar Data

The NASA-IBM Lunar Foundation Model analyzes 17 years of Moon imagery to accelerate crater mapping, ice detection, and volcanic feature identification.

Via AI Watch · Sep 10, 2026
AI· 3 min read

IBM and NASA Open-Source Lunar AI Model for Moon Exploration

The NASA-IBM Lunar Foundation Model synthesizes decades of multi-instrument observations to identify ice deposits, craters, and volcanic features with up to 23% better accuracy than existing methods.

Via AI Watch · Sep 10, 2026