Israel Bar Association Issues Strict Ethical Guidelines for AI Use in Legal Practice
The Israel Bar Association's National Ethics Committee has issued strict guidelines prohibiting autonomous AI agents from making legal decisions or drafting court filings without human oversight, warning lawyers against feeding confidential client data into open AI platforms.
The Israel Bar Association's National Ethics Committee has issued a strict directive prohibiting autonomous systems and AI agents from making substantial legal decisions, providing independent legal advice, or drafting court submissions without close human supervision. Signed by Adv. Menachem Moskowitz, chairman of the committee, the updated ethical opinion emphasizes that while artificial intelligence can serve as a controlled technical aid, legal and professional responsibility remains entirely with human attorneys. With approximately 100,000 lawyers currently practicing in Israel, the bar stresses that as AI autonomy increases, so do professional, ethical, and procedural risks.
Strict Boundaries on Autonomous AI Agents
According to the directive, an AI agent is defined as a system capable of acting with a certain degree of autonomy in executing tasks, planning workflows, and interacting with external environments without human intervention. The committee explicitly ruled that "the rule is that autonomous systems or AI agents must not be used in a manner that allows them to make substantive decisions regarding representation instead of the lawyer." Furthermore, AI cannot independently provide legal advice, represent clients before courts or third parties, submit legal pleadings, serve binding documents, or advance legal proceedings without sufficient human oversight and approval.
However, the Bar Association clarifies that using AI systems for technical, bounded, documented, and reversible tasks—driven by prior human decisions and subject to clear parameters and appropriate controls—is entirely permissible. The inclusion of an autonomous component does not inherently ban a system, but it demands heightened caution from lawyers to ensure suitability for representation.
Client Privacy and the Prohibition on Feeding Restricted Data
Protecting client confidentiality is a central pillar of the new guidelines. Attorneys are strictly prohibited from feeding "restricted data"—any non-public information protected by legal, professional, or ethical duties, such as medical secrets, sensitive personal details, or confidential business information—into open, unsecured AI platforms like public chatbots.
"The lawyer must avoid entering restricted data into open AI platforms or any other usage environment where there is insufficient certainty that the information will remain protected at a level consistent with professional duties," the committee stated.
The Bar warns that simple anonymization (removing names or identifiers) is insufficient protection, as advanced data cross-referencing can easily expose identities. Even when using closed platforms that guarantee data security and contractually prohibit using inputs for model training, attorneys retain a non-delegable duty of care to evaluate security protocols, encryption, and data retention policies.
Client Transparency and Managing AI Hallucinations
Transparency toward clients is required when AI use substantially shapes the representation goals, involves restricted data on open platforms, or significantly impacts professional decision-making. In cases where restricted data must be used in open platforms, explicit and informed client consent is mandatory. General or sweeping clauses in retainer agreements or powers of attorney allowing AI use are deemed insufficient.
Additionally, the committee issued a stern warning regarding AI "hallucinations"—instances where artificial intelligence fabricates entirely plausible yet completely false court rulings, citations, or legal sources. Citing previous cases where lawyers submitted fabricated precedents to courts, the Bar warned that blind reliance on computer outputs will result in severe sanctions, reputational damage, and disciplinary proceedings.
"The lawyer's personal and direct responsibility for actions, advice, legal documents, and pleadings remains absolute, and they may not hide behind AI-generated outputs," the directive concludes. Non-compliance with these guidelines will create a rebuttable presumption of ethical breach, while adherence serves as significant evidence of reasonable and ethical conduct.
