Fofanah v. Rockwall Rental: Texas Court Flags Hallucinated Citation in Appellants’ Reply Brief | Advocate Prakhar

Case at a Glance

Full Citation	Foday S. Fofanah and Hawa Fofanah v. Rockwall Rental Properties, No. 05-25-00536-CV (Tex. App. — Dallas, Apr. 15, 2026)
Court	Court of Appeals, Fifth District of Texas at Dallas
Date	April 15, 2026
Category	AI Hallucination in Appellate Brief
Jurisdiction	United States (Texas State)
AI Tool Used	Inferred AI (hallucinated citation from the court itself in reply brief)
Outcome/Sanction	Foreclosure judgment affirmed; no formal sanction imposed; court documented the hallucination

Background

Foday and Hawa Fofanah executed a promissory note in January 2020 secured by a deed of trust against their property in Kaufman County, Texas. They made one payment and then defaulted. The lender, Rockwall Rental Properties, authorised foreclosure, completed all required notices, and purchased the property at a public foreclosure sale in May 2024. The Fofanahs filed suit asserting wrongful foreclosure and related claims; the trial court granted summary judgment in favour of Rockwall Rental Properties, awarding possession, monetary damages, and attorney fees. The Fofanahs appealed, submitting opening and reply briefs that contained multiple citation problems, including what the appellate court identified as a hallucinated citation to a purported case from the Fifth District itself.

The AI Issue

The appellate court identified that the Fofanahs’ appellate briefing was deficient in two ways related to AI: first, citations in the opening brief did not support the propositions for which they were offered; second, the reply brief included fifteen citations, one of which was “a purported opinion from this court for which the citation is incorrect and which we were unable to locate after extensive research” that “appears to be hallucinated.” The description — a case attributed to the exact court hearing the appeal that does not exist — is a classic AI hallucination pattern where generative AI invents citations that plausibly fit the context of the relevant court and jurisdiction.

What the Court Decided

• The appeal was dismissed on the merits; the summary judgment in favour of Rockwall Rental Properties was affirmed [foreclosure upheld].
• The opening brief was characterised as a “disaster”: seven cited authorities supported none of the four appellate issues raised [zero citation accuracy].
• One citation in the reply brief was specifically identified as a hallucinated case: the court was unable to locate it after “extensive research” [hallucination documented on record].
• Four of the fifteen reply brief citations did not support the asserted propositions [pattern of citation unreliability].
• No formal sanctions were imposed, but the court’s documentation of the hallucination creates a permanent record of the briefing deficiencies [judicial record without monetary penalty].

“Appellants’ briefing is a disaster… [One citation in the reply brief] appears to be hallucinated… we were unable to locate [it] after extensive research.”

— Justice Lee, Fifth District Court of Appeals, Dallas, April 15, 2026

The India Angle

Indian Law Equivalent

In India, appellate briefs filed in High Courts and the Supreme Court are governed by the respective court’s rules on written submissions. The Supreme Court Rules, 2013 (Order XXII) and High Court Original Side Rules require accurate citation of authority. AI-hallucinated citations in Indian appellate briefs would expose advocates to professional misconduct proceedings under the Advocates Act, 1961, and could lead to orders under Order XI CPC imposing costs for wasted judicial time. The Contempt of Courts Act, 1971 could apply if the hallucination is found to be wilful.

Bar Council Rules

BCI Rule 14 (no false statements), Rule 15 (no disrepute), and Rule 22 (dignity of court) are directly engaged by AI-hallucinated citations in appellate briefs. In India, where appeals to High Courts and the Supreme Court are high-stakes and where judicial time is severely constrained, the harm from hallucinated citations in appellate proceedings is disproportionately large. The BCI should consider issuing specific guidance on AI use in appellate brief preparation.

Practical Advice for Indian Advocates

• In appellate practice, verify every citation using the court’s own official database or authenticated law reports before inclusion in written submissions — AI has a documented tendency to hallucinate cases attributed to the very court being addressed.
• When reviewing AI-generated drafts of appeal memoranda, pay particular attention to citations from the same court — hallucinated cases are most plausible when they mimic the jurisdiction and court level of real cases.
• File a correction affidavit immediately if you discover citation errors after submission — proactive correction demonstrates good faith and typically results in more lenient judicial treatment than errors discovered by the court or opposing counsel.

Quick Takeaways

• AI tools hallucinate cases attributed to the exact court hearing the appeal — the most dangerous pattern.
• Courts are independently researching citations; a hallucination will be found and documented.
• Zero citation accuracy in an appellate brief characterises the entire appeal as unreliable.

---

Deep Dive: Hallucinated Home-Court Citations — The Most Dangerous AI Error in Appellate Practice

The Fofanah case illustrates what practitioners and courts are increasingly recognising as the most dangerous specific form of AI hallucination in appellate practice: the fabrication of citations attributed to the very court that is hearing the appeal. When an AI tool generates a citation to a “Fifth District Court of Appeals” case that does not exist, it is producing text that is maximally plausible for the appellate court receiving it — the citation appears in the right format, references the right court, and matches the jurisdictional and subject-matter context of the actual appeal. The hallucination is hardest to detect precisely because it looks most legitimate.

The underlying foreclosure dispute in Fofanah was unexceptional. The Fofanahs defaulted on a 2020 promissory note, the lender completed all required statutory notices, purchased the property at public sale in May 2024, and obtained summary judgment. None of the procedural steps were challenged successfully on appeal. The case reached the appellate court because the borrowers filed suit asserting wrongful foreclosure claims that the trial court correctly rejected. What elevated the case to a teachable AI law moment was the quality — or rather the lack of quality — of the appellate briefing.

The Fifth District’s characterisation of the opening brief as “a disaster” is unusually blunt language for an appellate opinion. The court’s finding that all seven cited authorities in the opening brief supported none of the four appellate issues is statistically remarkable: a genuinely random selection of cases from the relevant legal area would produce at least some citation support by chance. The complete absence of accurate authority suggests that the citations were generated by an AI tool that was matching stylistic patterns rather than legal substance — producing cases that sounded relevant without being relevant.

For Indian appellate practitioners, the Fofanah pattern has direct implications. The Supreme Court of India and High Courts routinely receive written submissions from advocates who have used AI tools to generate case lists and research summaries. The hallucination risk is highest where advocates ask AI tools to suggest cases from a specific High Court or the Supreme Court on a specific legal point. The AI will confidently produce citations — some real, some fabricated — in the format and style of actual Indian case law. Without individual verification of each citation against authenticated databases, advocates are taking the same risk as the Fofanahs’ counsel — and the consequences in Indian appellate practice, where wasted judicial time can result in exemplary costs, are severe.