The Judiciary is moving to tighten oversight on the use of artificial intelligence in courts after a rise in cases where advocates submitted documents containing fake legal authorities and AI-generated material that failed to meet professional standards.
Draft policy documents circulated for public participation show that the Judiciary is developing fresh rules to guide how judges, magistrates, lawyers and court staff use AI technology in legal processes while protecting fairness, privacy and the independence of courts.
The move follows recent incidents where judges struck out pleadings after discovering that some of the cases and authorities cited by advocates were either fabricated or generated using AI tools.
According to the Judiciary, the proposed framework forms part of ongoing efforts to modernise court operations through technology while ensuring safeguards are put in place to prevent misuse.
“To support the practical implementation of the Policy, Draft Practice Directions have also been developed to guide judges, judicial officers, and court users,” the Law Society of Kenya (LSK) stated while inviting comments from members.
The proposed directions will require advocates to openly declare when AI tools have been used in preparing affidavits, submissions, witness statements, pleadings and other court documents filed before judges.
Lawyers using AI-assisted research will also be expected to personally confirm the accuracy of every case law, statute, regulation and legal authority before relying on it in court proceedings.
“An Advocate shall not file a document in court that has been wholly generated by an AI tool without meaningful human review, revision and professional judgment applied to its content. Wholesale adoption of AI-generated text without independent review shall be treated as a failure to exercise the professional judgment required of an officer of the court,” the draft directions state.
The Judiciary noted that many court systems across the world are already using AI tools to improve efficiency in areas such as legal research, document analysis, transcription services and scheduling of cases.
The draft policy says the technology could help ease the challenge of case backlog by automating routine administrative tasks and speeding up court processes.
AI-powered transcription systems could also help convert oral submissions into written records instantly, reducing workload for court staff and improving accuracy in court documentation.
Even so, the Judiciary warned that the growing use of AI carries risks that could affect justice delivery if not properly managed.
“The deployment of AI within a constitutional and human rights framework demands vigilance to ensure that technology enhances, rather than undermines, the principles of fairness, accountability, transparency and judicial independence,” the document states.
The draft policy highlights concerns around cybersecurity threats, protection of sensitive data and the possibility of bias from AI systems trained using historical information.
The Judiciary further stressed that AI should only assist judicial officers and must never replace human decision-making in court matters.
“The Judiciary shall maintain a clear boundary between administrative automation and adjudicative discretion, ensuring AI serves as a decision-support tool, not a decision-maker,” the draft policy states.
Under the proposed framework, judges and magistrates will still be required to exercise independent judgment and carefully verify all AI-generated content before relying on it for official duties.
The Judiciary also plans to restrict the use of third-party AI systems to administrative, drafting and research functions that do not involve confidential or personally identifiable judicial information.
“Any AI-generated content shall be treated solely as informational and shall be independently verified by a human user. Judicial officers and staff shall review and confirm the accuracy, context and reliability of all AI outputs before relying on them for official use,” the document adds.
The draft policy further directs judicial officers to rely on original legal sources, including statutes, judicial precedents and official law reports, instead of treating AI-generated material as primary legal authority.