Robinson v. Velosio LLC: Pro Se Plaintiff Sanctioned for AI-Hallucinated Cases in Employment Discrimination Brief | Advocate Prakhar

Case at a Glance

Background

Devin Robinson, a pro se plaintiff, sued his former employer Velosio LLC — one of the largest Microsoft Dynamics 365 implementation partners — alleging racial discrimination, retaliatory termination, hostile work environment, and violation of the Maryland Wage Payment and Collection Act. Robinson alleged he was fired ten weeks after being hired and that his final paycheck was unlawfully reduced for “overuse” of paid time off he had not yet accrued. He filed four versions of his complaint over the course of the litigation. When defendants moved to dismiss based on a settlement and release agreement Robinson had signed in exchange for two weeks severance pay, his response brief contained non-existent cases and fabricated quotations — prompting a separate sanctions motion.

The AI Issue

Defendants identified that Robinson’s response brief cited cases that did not exist and contained quotations that could not be verified. Robinson filed a “Declaration Regarding Citation Clarifications” attempting to explain the discrepancies, which strongly suggested AI assistance in drafting the brief. The court held a separate hearing on April 22, 2026 specifically to address the motion for sanctions — an unusual procedural step that underscores how seriously the court treated the citation fabrication issue, even in a pro se context where courts ordinarily apply liberal construction standards to filings.

What the Court Decided

• The court granted defendants’ motion to dismiss, holding that Robinson’s settlement and release agreement knowingly and voluntarily waived all discrimination claims [ADEA/Title VII waiver standard].
• The release was enforced despite Robinson’s duress argument because he failed to plead that defendants engaged in improper acts beyond economic pressure [duress standard in contract law].
• Robinson’s Maryland Wage Payment and Collection Act claim failed because any deduction was authorised by the employment handbook he had signed [statutory wage claim].
• The court conducted a sanctions hearing addressing the non-existent cases and fabricated quotations in the response brief [Rule 11 / inherent powers].
• After four versions of the complaint, the dismissal was granted with prejudice — Robinson could not demonstrate he could cure the fundamental defects [prejudice standard].

“Mr. Robinson also filed a ‘Declaration Regarding Citation Clarifications’ in which he attempts to explain why his response brief contains non-existent cases and quotations, as pointed out in Defendants’ reply brief. On April 22, 2026, the Court held a hearing on Defendants’ motion for sanctions, during which the Court provided the parties an opportunity to address the motion to dismiss.”

— U.S. District Court, District of Maryland, 5 May 2026

The India Angle

Indian Law Equivalent

India’s employment discrimination framework under the Constitution of India (Articles 14, 15, and 16) and sector-specific statutes like the Equal Remuneration Act, 1976 and Prevention, Protection and Redressal of Sexual Harassment of Women at Workplace Act, 2013 (POSH Act) provide the substantive backdrop for discrimination claims. The court’s power to sanction for false citations mirrors the inherent powers of Indian courts under Section 151 CPC and Order XI Rule 1 provisions on verification, as well as the Contempt of Courts Act, 1971.

Bar Council Rules

While Robinson was a pro se litigant, any Indian advocate who assists a client in drafting briefs and then allows AI-hallucinated citations to go unchecked into the filing would face consequences under Bar Council of India Rules, Part VI, Chapter II, Section II, Rules 1, 2, and 14 — specifically the prohibitions on misleading the court and misrepresenting the law. The “Declaration Regarding Citation Clarifications” filed by Robinson represents a post-hoc admission of AI use that aggravated rather than mitigated the court’s concern.

Practical Advice for Indian Advocates

• When assisting pro se clients draft pleadings or briefs that use AI tools, advocates must review and verify every citation before the document is filed — the liberal pleading standard for pro se filers does not extend to fabricated citations.
• Employment disputes in India increasingly involve documentary evidence (offer letters, employment handbooks, settlement agreements) — any AI-generated summary or analysis of such documents must be verified against the originals before reliance.
• A post-hoc “clarification” declaration about AI-generated citation errors may itself create an adverse inference — honesty upfront and thorough pre-filing review are always preferable to explanations filed after the error is exposed.

Quick Takeaways

• Liberal construction for pro se filers does not insulate litigants from sanctions for AI-hallucinated citations.
• A separate sanctions hearing for citation fabrication adds procedural cost and reputational harm even when the underlying claim fails on merits.
• Settlement releases with clear waiver language will be enforced even when the signatory later claims AI-assisted confusion about the law.

Deep Dive: The Pro Se AI Litigant — Access to Justice or Access to Fabrication?

Robinson v. Velosio sits at the intersection of two urgent debates: the AI hallucination crisis in legal filings, and the access-to-justice gap for self-represented litigants. Robinson’s situation is far from unique — across the US court system in 2025 and 2026, there has been an explosion of pro se filings that contain citations to cases that do not exist, produced by litigants who turned to free AI tools because they could not afford attorneys. The question these cases raise — whether courts should treat AI-assisted pro se filers differently from AI-assisted attorneys when sanctioning citation fabrication — remains deeply contested.

In Robinson’s case, the court took the unusual step of holding a dedicated sanctions hearing before ruling on the motion to dismiss, suggesting the citation issues were treated as a serious independent matter rather than an incidental curiosity. The filing of a “Declaration Regarding Citation Clarifications” — essentially an admission that the brief contained AI-generated content that Robinson could not independently verify — created an evidential record that courts across the country are increasingly seeing: litigants who acknowledge AI use after the fact, in the face of challenge, rather than disclosing it proactively and verifying independently before filing.

The employment law backdrop of the case adds another layer of complexity. Robinson’s underlying claims — racial discrimination, retaliation, hostile work environment — are serious civil rights allegations. The fact that all his claims were dismissed not on the merits but because of a settlement release he signed under financial pressure (two weeks severance) raises questions about whether AI assistance in litigation might paradoxically harm disadvantaged litigants by giving them a false sense of legal competence. A litigant who uses AI to research whether a settlement release can be challenged for duress, and receives a confident AI response citing non-existent cases supporting that position, is worse off than a litigant who approaches an attorney or a legal aid clinic for guidance.

For Indian advocates and courts, Robinson v. Velosio is a cautionary tale about the democratisation of legal tools. India has a large and growing population of litigants in person — particularly in consumer courts, labour tribunals, and family courts — who now have access to AI-powered legal assistants in Hindi, Tamil, and other regional languages. The risk that these tools will hallucinate citations to Indian statutes, rules, and case law is real and growing. Indian district courts and subordinate courts, which currently lack robust citation-verification infrastructure, are particularly vulnerable to this problem. Bar associations and legal aid societies in India should consider proactive outreach to educate litigants in person about the risks of using AI-generated legal citations without independent verification.