Ella and Ramin LLC v. Travelers Insurance: AI Use by Pro Se Litigant Triggers Judicial Warning | Advocate Prakhar

Case at a Glance

Background

Ramin Ghayoori, the owner of a condominium unit in Chicago, suffered substantial water damage from recurring roof leaks beginning in September 2021. He submitted multiple claims to TravCo Insurance Company, which either denied them outright or offered what he considered inadequate compensation. Ghayoori, dissatisfied with the insurer’s remote-only assessment and its reliance on an inaccurate condo association report, filed a pro se lawsuit in federal court in the Northern District of Illinois.

Ghayoori named not only TravCo as defendant, but also Travelers Insurance Company (asserting it was TravCo’s parent) and Alan Schnitzer, purported CEO of the parent company, who had ignored multiple emails from Ghayoori. His amended complaint raised five claims: breach of contract, bad faith, negligent infliction of emotional distress (NIED), intentional infliction of emotional distress (IIED), and uninhabitable living conditions. TravCo moved to dismiss Counts II through V.

During motion practice, defense counsel noticed that Ghayoori’s response brief contained significant legal errors and suggested these appeared to result from unchecked AI use. Ghayoori fired back with a motion for Rule 11 sanctions against defense counsel, arguing that pointing out AI-generated errors was itself sanctionable conduct. The court disagreed emphatically.

The AI Issue

The central AI-related question was whether a pro se litigant who admits using generative AI can escape responsibility for the errors that AI introduces. Ghayoori openly acknowledged using AI to prepare his filings. Yet the court found his briefs contained egregious citation failures: a case cited as establishing that parent companies and subsidiaries may be treated as one entity (Harris v. One Hope United, Inc.) turned out to be a case about a discovery privilege — completely unrelated to corporate law. More strikingly, a case cited repeatedly (Heritage Common Partners v. Village of Summit) could not be found at either of the two reporter citations Ghayoori provided; the only cases with that name in federal court were from 1990 and 1991 and related to entirely different topics.

What the Court Decided

• Defense counsel’s act of flagging AI-generated legal errors to the court was not sanctionable — it was an obligation of officers of the court.
• Plaintiff’s motion for Rule 11 sanctions was denied as frivolous.
• Counts III (NIED) and IV (IIED) were dismissed without prejudice for failure to allege contemporaneous physical injury and extreme-enough conduct respectively.
• Count V (uninhabitable conditions) was dismissed with prejudice as not an independent tort under Illinois law.
• Travelers Insurance Company and CEO Schnitzer were dismissed as defendants with prejudice — they were not parties to the insurance contract.
• Count II survived, recharacterized as a claim under Section 155 of the Illinois Insurance Code for vexatious and unreasonable conduct.

“[A]ll litigants — represented and unrepresented — must read their filings and take reasonable care to avoid misrepresentations, factual and legal.”

— Judge Andrea R. Wood, N.D. Illinois, March 31, 2026 (quoting Seventh Circuit in Jones v. Kankakee Cnty. Sheriff’s Dep’t)

The India Angle

Indian Law Equivalent

In India, an insurance policyholder’s equivalent remedy for bad-faith claim denials lies primarily under the Insurance Act, 1938 (as amended), and through complaints to the Insurance Regulatory and Development Authority of India (IRDAI). The Consumer Protection Act, 2019 also provides a strong avenue: Section 2(11) defines “deficiency” broadly to include inadequate investigation and unreasonable denial of claims, and the National Consumer Disputes Redressal Commission (NCDRC) has consistently held that insurers who rely on flawed third-party reports without independent verification commit deficiency in service. Courts may award compensation for mental agony caused by insurer misconduct under Section 39 of the Consumer Protection Act.

Bar Council Rules

Bar Council of India Rule 49 (under Chapter II, Part VI of the BCI Rules) requires every advocate to “make no misstatement of law or facts to the court or any tribunal.” Rule 52 further prohibits an advocate from knowingly relying on fabricated or non-existent authorities. These obligations apply regardless of the medium — including AI — used to prepare submissions. An advocate who files AI-generated content without verification risks professional misconduct proceedings.

Practical Advice for Indian Advocates

• Treat every AI-generated citation as an unverified lead: cross-check the reporter citation, read the headnote and relevant paragraph, and confirm the case stands for exactly the proposition you assert.
• In insurance disputes, the ground of “vexatious delay” under Indian law mirrors Section 155 of the Illinois Insurance Code — document every communication with the insurer meticulously to build a timeline of bad faith.
• If opposing counsel or the court flags citation errors, acknowledge and correct them promptly; attempting to counter such corrections with cross-sanctions motions, as Ghayoori did, will likely be seen as frivolous and damage your credibility.

Quick Takeaways

• Pro se litigants who use AI accept full responsibility for citation accuracy — ignorance of AI errors is no defence.
• Defense counsel who flag AI hallucinations to a court are fulfilling a professional duty, not committing a sanctionable act.
• An insurer’s vexatious refusal to investigate claims properly may support a statutory bad-faith remedy even if a freestanding “bad faith” tort is not recognised in the jurisdiction.

Deep Dive: The Duty of Verification When Using AI in Legal Drafting

The Ella and Ramin case is a microcosm of a broader phenomenon sweeping courts across the United States and increasingly appearing in common-law jurisdictions worldwide: the phenomenon of litigants — both represented and unrepresented — submitting AI-generated legal briefs without any independent verification of the cases cited. What makes this case particularly instructive is that the AI errors here were not subtle. Ghayoori cited a discovery-privilege case for a corporate-law principle, and he cited a case whose very coordinates — the year, the reporter, the volume and page numbers — led nowhere. Any five-minute Westlaw or Google Scholar check would have exposed both errors immediately.

The legal-ethics dimension is stark. When a lawyer or pro se litigant signs a pleading or brief, they certify under Rule 11 of the Federal Rules of Civil Procedure (or its state equivalents) that the legal contentions therein are “warranted by existing law.” Generative AI systems are designed to produce fluent, confident-sounding legal prose. They are not designed to be accurate. They “hallucinate” — a technical term meaning they generate plausible-sounding but factually incorrect information — with alarming frequency in the legal domain, because legal citations require precision (exact parties, volume, page, year, court) that AI language models are prone to confabulate.

The Seventh Circuit’s language quoted by the district court — that all litigants “must read their filings and take reasonable care to avoid misrepresentations” — establishes a technology-neutral standard. The obligation predates AI by decades, but the AI era has made compliance far more demanding. A lawyer who once drafted a brief from primary research was implicitly verifying citations as they wrote. A lawyer who pastes AI output into a brief and edits for style is no longer doing primary verification; they have outsourced research to a tool that invents authorities.

The practical solution courts implicitly demand — and that the Seventh Circuit was moving toward in the Jones v. Kankakee case referenced in this opinion — is a two-step verification protocol: first, confirm the case exists (Westlaw, Lexis, Google Scholar); second, read the relevant passage and confirm it supports the exact proposition for which you are citing it. This sounds obvious. But surveys of practitioners reveal that under time pressure, many lawyers skip step two, trusting AI to have captured the holding correctly. As Ella and Ramin shows, that trust is misplaced.