Israeli Bar Association Restricts AI Use by Lawyers

The National Ethics Committee of the Israel Bar Association has issued a new directive prohibiting lawyers from allowing autonomous systems or AI agents to make significant legal decisions, provide independent legal advice, draft legal documents, or advance legal proceedings without close human supervision and approval. The committee clarified that while AI can serve as a controlled technical aid, the ultimate legal and professional responsibility remains solely with the human lawyer, emphasizing that human judgment is irreplaceable.

The directive addresses the increasing autonomy of AI systems, known as AI Agents, which can operate with less direct human intervention. Lawyers are warned that as AI autonomy increases, so do the ethical, professional, and procedural risks. The committee stressed that AI agents are systems capable of acting with a degree of autonomy in tasks, planning workflows, selecting tools, and interacting with external systems without human input.

While AI use for technical, bounded, and reversible actions based on prior human decisions is permissible under strict controls and supervision, lawyers must carefully assess the suitability of AI tools for representation, the nature of actions the AI can perform, and the level of human oversight. The core principle is that lawyers bear personal and direct responsibility for all professional actions, decisions, advice, arguments, or outputs generated using AI. This includes safeguarding client information, confidential data, and personal details, and avoiding any use that could lead to unauthorized disclosure, privacy breaches, or violations of confidentiality.

Furthermore, lawyers are cautioned against inputting "limited information", non-public data subject to legal or ethical confidentiality obligations, such as medical secrets or sensitive client details, into unsecured, open AI systems like public chatbots. Even simple anonymization techniques are deemed insufficient, as data cross-referencing could potentially reveal identities. The directive also mandates that lawyers must obtain informed and explicit client consent before inputting limited information into open AI platforms, especially if anonymization is not fully effective. General clauses in retainer agreements allowing AI use are not considered sufficient consent.

The committee also highlighted the risk of AI "hallucinations," where AI systems fabricate plausible but incorrect legal citations, rulings, or sources. Blind reliance on AI outputs, especially those containing fabricated information, can lead to sanctions, reputational damage, and disciplinary action. Lawyers are urged to understand that AI systems often generate statistically "correct" or convincing responses that may be factually or legally inaccurate, incomplete, biased, or based on false premises. While AI can offer insights, drafting assistance, and initial work, it cannot replace human professional judgment, factual investigation, and decision-making. Failure to comply with the directive may create a presumption of ethical misconduct.