Every few decades, a technology arrives that changes how legal work is done — the photocopier, the online research database, e-mail. Artificial intelligence is the latest, and in many ways the most powerful. But a lawyer's professional obligations do not change with the tools. Whatever the software does, the duties of confidentiality, competence, and candour remain personal to the lawyer. Here is how we think about the ethics of AI in practice.
Confidentiality comes first
Client information is not raw material. Before any AI tool touches a client document, a firm needs clear answers: Where is the data processed and stored? Is it retained? Is it used to train models that serve other customers? Can the provider's staff see it? If the answers are unclear, the tool has no business near client files. The duty of confidentiality is not satisfied by a vendor's marketing page — it requires understanding, in concrete terms, what happens to the information.
Accuracy is a duty, not a feature
Generative AI systems produce fluent, confident text that is sometimes wrong — including, notoriously, citations to court decisions that do not exist. Courts in multiple jurisdictions have publicly rebuked and sanctioned lawyers for filing AI-fabricated authorities. The lesson is not "never use AI"; it is that verification is non-negotiable. Every authority checked against the official source. Every factual claim traced to the record. Competence in 2026 includes knowing precisely what these tools get wrong.
Software can draft; only a lawyer can be accountable. Responsibility for advice cannot be outsourced to a model.
Accountability and honesty with clients
Clients are entitled to straight answers about how their matter is handled. In our view, that includes technology. Three commitments follow:
- A lawyer owns every deliverable. Nothing generated by software reaches a client or a court without review and judgment by a person whose name stands behind it.
- Efficiency should show up in the bill. If tools make routine work faster, transparent billing means the client sees that benefit — not a technology premium.
- Questions get real answers. Ask us how we use technology on your matter and you will get a specific, plain-language explanation, in English or Spanish.
Fairness and the human element
There is a quieter ethical issue beneath the headlines: AI systems learn from historical data, and historical data carries historical bias. In a profession whose entire purpose is fair treatment, outputs that touch people — settlement ranges, risk assessments, even tone of correspondence — deserve human scrutiny for fairness, not just accuracy. Disputes and family transitions are human events. The measure of any tool is whether it helps a lawyer serve the human being in front of them better: faster answers, clearer costs, steadier guidance. That is the standard we hold technology to — and the standard clients should hold every firm to.