Build meeting-level privilege controls into AI recaps

On September 30, the New York City Bar Association issued an ethics opinion cautioning attorneys not to record nonclient conversations with artificial intelligence tools, according to Law360 Legal Industry. The opinion reflects an emerging view that AI meeting recaps are a distinct business record category, Law360 Legal Industry reports, and says counsel should set controls at the meeting level and update them at the framework level. William Wright at Faegre Drinker is cited in the report. The immediate consequence is not a general prohibition on AI meeting tools; it is a demand that their recording, retention and use be handled as part of legal work rather than as neutral office software.

That matters because privilege handling is part of the product design for an AI-native legal service. A system that turns conversations into recaps does more than save time: it creates a business record whose treatment can affect discovery and confidentiality. The New York City Bar Association opinion, as reported by Law360 Legal Industry, therefore moves the question from whether an AI tool is secure to how the tool behaves in a particular meeting and what happens to the resulting record. Legaltech News (Law.com) reports that uncertainty about how AI fits within legal and discovery paradigms will likely continue for years despite more guidance and case law. That uncertainty is not a reason to treat privilege as an afterthought; it is the reason the controls have to be designed into the service.

The opinion does not establish that every AI meeting recap is unprivileged, that every recording of a nonclient conversation is improper, or that a legal service loses protection merely because AI is involved. It establishes a narrower point: the purpose of the conversation, who is present, whether the conversation is recorded and how the resulting information is handled can become part of the legal analysis. The distinction also fits Legal Futures’ account that AI can perform the work but cannot inherit the lawyer’s duty, leaving lawyers personally responsible for judgments they may no longer have independently reasoned through. A stronger reading would require a categorical rule from the New York City Bar Association or a court that AI meeting recaps receive one fixed treatment. The cited opinion instead points toward meeting-level decisions and framework-level updates. Neither AI destroys privilege nor a legal chatbot makes everything privileged is an adequate operating assumption.

The next confirming facts are concrete: a later ethics opinion or court ruling addressing whether an AI meeting recap must be produced in discovery; a firm policy specifying when such tools may record nonclient conversations; and product documentation showing meeting-level controls for recording, retention and access. A result that treats every recap the same would cut against the design-based reading. A rule or policy that varies by meeting purpose, participants and data handling would strengthen it. We will count the controls and the resulting treatment of records, not the presence of an AI label, as the test of whether the service has taken privilege seriously.