Missouri AI Ethics Rules for Lawyers

The official instruments

An opinion that began as a policy question

Missouri’s AI instrument has the most practical origin story in the dataset: a lawyer sat down to design a firm AI-use policy, asked the Missouri Office of Legal Ethics Counsel for guidance, and Informal Opinion 2024-11 came back on April 25, 2024. The mechanism matters as much as the content. Missouri Supreme Court Rule 5.30(c) provides for official informal advisory opinions, so 2024-11 is not a blog post or a committee article; it is the state’s designated ethics office answering through its designated channel. It is also, by that same mechanism, informal rather than formal, and this tracker types it as guidance for exactly that reason. Missouri belongs on the map with an honest label: real authority, informal tier.

Because the question was “how do I build a policy,” the answer doubles as a policy outline. Firms elsewhere get opinions organized by rule number and have to translate them into operations; Missouri firms get operations directly.

Five instructions, ordered by when they bite

Read as a sequence, the opinion’s instructions attach to different moments in a tool’s life. Before adoption: get training, so the people bringing AI into the practice understand what they are adopting. Also before any client data flows: assess the platform’s terms and security, the vetting step that decides whether client information can enter the tool at all; the risk-tier framework offers one way to structure that assessment. During use: verify AI-assisted content the way nonlawyer work is verified, importing a supervision standard every firm already runs, and preserve independent professional judgment, which is the opinion’s most distinctive entry. Verification catches false output; the independence duty addresses something subtler, the drift toward letting the tool’s answer become the analysis. Missouri states them separately because passing a citation check does not prove a lawyer actually exercised judgment. Finally, at billing: consider fee reasonableness, keeping AI’s economics inside the ordinary reasonableness analysis; the billing explainer collects how other instruments handle the same question with more specificity.

The nonlawyer-work analogy deserves a beat of attention, because it quietly assigns responsibility. Work product from a paralegal or law clerk goes out under a lawyer who reviewed it and owns it. Slotting AI into that same review lane means AI output never carries its own authority; someone with a license adopts it or it goes nowhere.

From answer back to policy

The opinion closes the loop it opened: firms should build an ethical framework and train both lawyers and nonlawyers. Note the both. An AI policy that trains associates while support staff paste client facts into consumer chatbots misses where much day-to-day risk actually lives, so the training obligation deliberately spans the whole office. For the framework itself, the elements the opinion names, training records, platform assessments, verification checkpoints, judgment-preservation norms, and fee review, map one-to-one onto the policy template, and the broader survey of what bars require in firm AI policies shows Missouri asking for the same core artifacts as its neighbors, just phrased as advice to a policy drafter.

Two boundaries complete the picture. Statewide, our dataset records no Missouri court rule on generative AI in filings, so courtroom obligations arrive judge by judge through standing orders, which get confirmed before each filing per the court orders guide. And temporally, an informal opinion is a first word rather than a last one; if Missouri ever escalates to a formal instrument, the tracker will reflect it and the changelog will date the change.