A troubling pattern has emerged in legal proceedings across the United States and globally: attorneys are submitting court documents generated by artificial intelligence chatbots—and facing serious repercussions when these materials prove unreliable. Earlier this year, a US judge discovered entirely fabricated case citations embedded in a lawyer's legal brief, a revelation that led the attorney to acknowledge using Claude, an AI system, to compose the filing. This incident exemplifies a broader challenge confronting the legal profession as technology infiltrates courtrooms faster than professional standards can adapt.
The problem extends far beyond a single errant attorney or one questionable document. Judges across numerous jurisdictions have begun encountering briefs peppered with false legal citations, invented case names, and distorted representations of established precedent—all hallmarks of what AI researchers term "hallucinations," where language models fabricate plausible-sounding but entirely inaccurate information. What makes this particularly alarming for the judiciary is that these errors often appear credible at first glance, requiring time-consuming fact-checking to uncover. For a court system already straining under heavy caseloads, this represents an unwelcome additional burden.
The tension between innovation and accountability is becoming increasingly acute. Artificial intelligence tools like Claude, ChatGPT, and Google's Gemini have sophisticated capabilities that make legal research and document drafting genuinely useful—when deployed responsibly. These systems can rapidly synthesize case law, suggest logical arguments, and identify overlooked statutory provisions. However, they fundamentally lack the precision and accountability that the legal system demands. Unlike traditional legal research databases such as Westlaw or LexisNexis, which are explicitly designed for accuracy and cite specific sources, generative AI systems operate by predicting statistically likely text sequences rather than verifying information against authoritative sources.
Professional responsibility rules governing attorneys have historically emphasised their duty to ensure all submissions to courts are accurate and truthful. Many bar associations, including those in the United States, Canada, and increasingly the United Kingdom, have issued guidance cautioning lawyers against using AI tools without rigorous human review and verification. Yet the guidance has proven insufficient. Some attorneys, attempting to streamline workflows or reduce costs, have skipped essential verification steps or failed to comprehend the limitations of AI systems. This negligence has directly undermined judicial integrity.
For Southeast Asia and Malaysia specifically, these developments carry particular significance. While the Malaysian legal profession has not yet reported major incidents involving AI-generated briefs, the region's courts will almost certainly face similar challenges as technology adoption accelerates. Malaysian lawyers increasingly use digital tools in research and writing, and generative AI platforms are becoming more accessible and affordable. Without clear professional standards and enforcement mechanisms now, Malaysian courts could find themselves confronting the same credibility problems that have prompted disciplinary action in Western jurisdictions.
Several high-profile cases have crystallised judicial disapproval. In one notable instance, a New York attorney submitted a brief containing multiple non-existent cases, each with fabricated judicial decisions. When questioned, he acknowledged using ChatGPT without adequately verifying the outputs. The court imposed sanctions, expressing clear frustration that such fabrications had wasted judicial resources. Other jurisdictions have similarly punished attorneys for submitting AI-generated documents marred by errors or omissions, with some facing disciplinary proceedings that could threaten their license to practice.
The problem is further complicated by attribution and transparency challenges. Many attorneys who quietly use AI tools for initial drafts may not explicitly disclose this to courts. The question of whether lawyers must inform judges that they employed AI assistance remains unsettled across most jurisdictions. Some courts are beginning to require such disclosure, arguing that judges deserve to know when documents have been created using automated systems of uncertain reliability. Others maintain that the final product—its accuracy and quality—matters more than the tool used to generate an initial draft.
Bar associations and legal regulators are now scrambling to establish clearer standards. The American Bar Association and comparable bodies elsewhere have begun developing specific protocols for responsible AI use in legal practice. These guidelines typically insist that attorneys remain ultimately responsible for verifying all information in submitted documents, understand the limitations of AI tools, and consider whether AI use is appropriate for the particular task at hand. However, translating these principles into enforceable rules that protect the integrity of legal proceedings without stifling beneficial technological innovation remains a delicate balancing act.
Legal educators are also responding by incorporating AI literacy into law school curricula, ensuring future practitioners understand both the capabilities and profound limitations of these systems. Some universities now teach students how to use AI as a starting point for research or writing whilst emphasising the non-negotiable requirement for human expert verification. This educational evolution recognises that AI tools are unlikely to disappear from legal practice—instead, the challenge is ensuring they are wielded competently and ethically.
For Malaysian courts and the legal profession, the international experience offers clear lessons. Establishing robust professional standards governing AI use before problems proliferate is far preferable to reactive disciplinary action afterwards. Bar councils and judiciary bodies should work collaboratively to clarify expectations, provide training, and enforce accountability. Clear rules about disclosure, mandatory verification procedures, and consequences for negligence will help maintain public confidence in the legal system whilst still permitting beneficial use of technological tools.
The broader implication is that artificial intelligence, for all its promise, cannot replace human professional judgment and accountability in domains where truth and accuracy fundamentally matter. Courts exist to dispense justice based on law and facts. When lawyers submit documents containing fabricated information—whether through negligence, ignorance, or intent—they undermine the entire system. As judges worldwide draw a firm line against AI-generated inaccuracy, they are effectively reasserting a timeless principle: the integrity of legal proceedings depends ultimately on human responsibility.
