The widespread adoption of generative artificial intelligence (AI) systems has revolutionized information processing, drafting, and research, enhancing efficiency and saving time. Professionals in the legal sector are increasingly utilizing AI tools for legal research, drafting, and case preparation on a global scale. However, a significant concern arises from the potential for generative AI to produce inaccurate citations. Errors in legal decisions based on non-existent or misrepresented legal authorities could lead to miscarriages of justice and undermine public trust in the legal system.
Within the legal domain, these “hallucinations” may manifest as fictitious case laws, fabricated party names, incorrect citations, misquoted judgments, and inaccurately stated statutory provisions. While the benefits of AI are acknowledged, the question of accountability arises in the event of a legal error caused by AI. The responsibility typically falls on the human actor presenting the information to the court or relying on it in legal reasoning, as AI is a tool and not a legal professional.
As of August 2026, Damien Charlotin’s AI Hallucination Cases Database has documented approximately 1,934 incidents globally. The 2023 Mata v Avianca, Inc. case marked the initial occurrence where fictitious cases generated by ChatGPT were cited in a US court filing, resulting in a $5,000 fine for the attorney involved. Subsequent instances of AI-generated citations have surfaced in various jurisdictions worldwide, including the United Kingdom, Australia, France, Canada, and India.
In the recent Pooja Ramesh Singh v Jammu and Kashmir Bank Ltd & Anr case in July 2026, the Indian Supreme Court emphasized that any decision based on false citations, even partially, holds no legal validity. The court urged the Bar Council of India to establish disciplinary rules for advocates citing AI-generated false authorities. This landmark ruling establishes a zero-tolerance approach in India towards fabricated citations, reinforcing the judiciary’s commitment to accuracy and accountability. Additionally, the Kerala and Gujarat high courts have introduced an AI policy for judicial officers to regulate the responsible and limited application of AI in legal processes.
In Bangladesh, the ICT Division has drafted an AI policy to promote responsible AI development. The Cabinet Division has issued directives cautioning against inputting sensitive data into AI systems, with Bangladesh Bank also prohibiting the inclusion of confidential banking information to prevent potential breaches. Chief Justice Zubayer Rahman Chowdhury has underscored the importance of defining the boundaries for AI usage in drafting court judgments, advocating for clear laws and policies in this domain. These initiatives demonstrate a growing acknowledgment of both the benefits and risks associated with generative AI across public institutions in Bangladesh.
As the integration of AI into the legal sector becomes inevitable, there is a pressing need for a practical directive on AI usage in the judiciary. Challenges include the low level of AI literacy among justice sector stakeholders and the quality and accessibility of digital legal databases. Addressing these challenges requires the formulation of a judicial AI-use policy based on global experiences, outlining restricted areas for AI deployment and enforcing penalties for misuse of AI-generated materials in judicial proceedings. The Bangladesh Bar Council should establish guidelines for lawyers to verify AI-generated citations and clarify accountability for false legal references. Furthermore, educating judges, lawyers, and court personnel on AI utilization is essential, along with creating a reliable digital legal database for verifying AI-generated information independently.
With proper regulation and guidance, AI can serve as a valuable tool to enhance legal processes, alleviate case backlog, and support various administrative functions. Bangladesh should consider developing a tailored judicial AI policy that aligns with its legal, institutional, and technological landscape to leverage the benefits of AI effectively.
