Generative AI tools are creeping into legal work in Bangladesh, from legal research to drafting and case preparation. But the technology invents fake citations, and the judiciary has no rules yet for who answers when a fabricated case lands in a court filing.
"If a legal decision is based on a non-existent or inaccurately represented authority, this could result in miscarriage of justice and erode public confidence in the judicial system," says Md. Ziaur Rahman, an MPhil fellow at the Institute of Bangladesh Studies, University of Rajshahi.
The problem is global. Damien Charlotin's AI Hallucination Cases Database has recorded some 1,934 incidents globally as of August 2026. The first recorded case, Mata v Avianca, Inc. in 2023, saw a US court fine an attorney $5,000 for citing ChatGPT-generated fake cases. Since then, similar incidents have surfaced in the UK, Australia, France, Canada, and India.
In July 2026, India's Supreme Court ruled in Pooja Ramesh Singh v. Jammu and Kashmir Bank Ltd & Anr that an order based on fake citations is "no decision in the eyes of law." The court directed the Bar Council of India to frame disciplinary rules for advocates who cite AI-fabricated authorities.
Bangladesh's courts are not prepared
Several Bangladeshi institutions have issued scattered warnings. The Cabinet Division cautioned against putting sensitive information into AI. Bangladesh Bank barred confidential banking data from AI systems. Chief Justice Zubayer Rahman Chowdhury has warned that AI must not write court judgments and called for laws defining where the technology can be used.
But the judiciary itself lacks a thumb rule. Rahman identifies three core problems: low AI literacy among judges, lawyers, and court staff remains rare; government legal databases are not fully digitized or indexed, raising the risk of fabricated citations; and the courts lack any mechanism to verify legal authorities before they are cited.
"AI use is inevitable and will enter the judicial arena," Rahman said. The question is whether the court system will set limits before an AI-produced error becomes a preventing order.
What a judicial AI policy should look like
Rahman's proposal is a judicial AI-use policy built on international experience. It should define where AI use is restricted or prohibited, and prescribe sanctions for those who misuse AI material in court proceedings. He urges the Bangladesh Bar Council to frame rules requiring lawyers to verify AI-generated citations, and for those presenting false authorities.
The policy should also include education and training for judges, lawyers, and court staff; a reliable, up-to-date digital legal database that allows independent cross-checking; and transparent disclosure whenever AI assisted in preparing petitions or submissions.
Rahman argues that a cited judgment can also reduce the case backlog plaguing Bangladeshi courts. "If properly guided and regulated, AI can become an innovative assistive tool to support research, case management, drafting, and other administrative functions," he said.
Why this matters for legal professionals
The judgment in India is a roadmap for what may come in Bangladesh. Attorneys who file business briefs or petitions built on AI-researched authorities risk professional discipline if those citations turn out to be fictional. Until a local policy exists, the practical safeguard is simple: run every AI-recovered case through a primary-source database before submitting it. Professionals who want structured training on legal research and AI compliance can start with an AI for Paralegals tutorial or explore broader AI for Legal resources.
Your membership also unlocks:
