Generative artificial intelligence has become a common tool in law firms, even as the rules governing its use are still being developed. Tasks such as drafting letters, summarizing cases, and preparing for meetings are increasingly handled with AI, often without clear guidelines. To address this gap, the Commission on Rules and Practices of the National Council of Bars has issued a guide titled Professional Ethics and Artificial Intelligence, offering practical advice for lawyers on how to use AI responsibly. The guide does not reject the technology outright but emphasizes that the core ethical principles of the legal profession must still apply. The guide clearly states that lawyers must never share client data with generative AI tools. This includes names, addresses, financial details, and any information that could identify a case. To protect this information, the guide recommends pseudonymizing or anonymizing data before use. Pseudonymization replaces identifying details with aliases that can be reversed, while anonymization removes the ability to identify individuals entirely. However, the guide warns that even masked data could be traced back to a person if combined with other details, such as rare functions or specific locations. The guide outlines three main criteria for selecting AI tools: sovereignty, security, and reliability. Lawyers are encouraged to use tools hosted in France or the European Union to avoid issues with foreign laws. Security measures, such as encryption and data deletion, are also emphasized. Reliability involves using AI that cites its sources and aligns with French legal standards. The guide cautions that even if a tool claims not to reuse data, this claim is often unverifiable. It also advises against using free AI tools that do not guarantee the confidentiality of legal work. The guide stresses that lawyers must always critically review and verify AI-generated content. Recent legal cases have shown that AI can fabricate references to non-existent case law, leading to potential liability. Lawyers are reminded that AI should never replace human judgment, and every reference, date, and figure must be checked manually. When using AI, lawyers must also inform clients about its role, especially if it involves their data or if they are interacting with an AI instead of a human. A standard clause can be added to contracts to clarify that AI is used as a technical aid, not as a substitute for human analysis. The guide recommends creating an internal AI usage policy, ensuring all staff are aware of best practices, and documenting procedures. It also warns that combining data from multiple cases in one AI session can create conflicts of interest, so each case should be handled separately. A simple one-page charter can outline approved tools, prohibited data, authorized uses, and procedures for reporting errors. The guide concludes with ten key reminders, such as never inputting identifiable data directly, always verifying AI-generated content, and using AI in a way that is fair and proportional. Ultimately, the guide presents a balanced approach, allowing AI to enhance efficiency while safeguarding client confidentiality and the integrity of legal practice.