Conversations about AI in law firms tend to swing between hype and alarm. Partners hear that AI will replace associates, or that any use is malpractice waiting to happen. Both extremes get in the way of sensible decisions. Here are seven common beliefs and what is closer to the truth, drawing on ABA Formal Opinion 512 (July 2024) and general practice. This is general information, not legal advice, and your state bar may have its own guidance.
Myth 1: "Anything I type into an AI tool stays private."
Reality: It depends entirely on the product and the plan. Some consumer tools may retain inputs and use them to improve models unless settings or contracts say otherwise. Business plans often offer stronger commitments, but you have to read the terms. Opinion 512 discusses the confidentiality risks, including that informed client consent may be needed in some situations.
Myth 2: "AI is accurate enough that I do not need to check it."
Reality: Generative AI can produce fluent, confident, completely invented content, including fake case citations. Courts have sanctioned lawyers for submitting such material. Competence under Model Rule 1.1 includes understanding a tool's limits and verifying output before relying on it.
Myth 3: "If we ban AI, we are safe."
Reality: A ban without an alternative often pushes use underground onto personal accounts, where the firm has no visibility. Many everyday tools now include AI features, so a ban may be hard to define or enforce. A better approach is an approved tool, clear rules and training.
Myth 4: "Only the lawyer who uses the tool is responsible."
Reality: Model Rules 5.1 and 5.3 address supervisory responsibilities. Partners and managers are expected to make reasonable efforts to ensure that those they supervise use technology appropriately. A firm-level policy is part of that effort.
Myth 5: "AI saves so much time that billing does not change."
Reality: Opinion 512 discusses fees, including that lawyers should bill honestly and that they generally may not charge clients for time they did not spend. How to handle flat fees versus hourly billing, and charges for tool costs, depends on the engagement agreement and your jurisdiction. Discuss it with ethics counsel and with clients up front.
Myth 6: "Our IT provider will make sure AI use is safe."
Reality: IT can configure access, review vendors and block risky tools, but decisions about how lawyers use AI in client work are professional judgments. Technical and legal staff need to work together. IT supports; lawyers decide what is appropriate and verify the work.
Myth 7: "AI tools are only for big firms."
Reality: Small firms may benefit significantly from tools that help with drafting, summarizing and organizing, since they have fewer people to absorb routine work. The same diligence applies at any size, but a solo or small firm can adopt a single carefully chosen tool with a one-page policy.
What a sensible middle path looks like
- Inventory current use. Ask staff, without blame, what they already use.
- Choose one or two approved tools after a vendor review covering data use, retention, access control and contracts.
- Write a short policy. Define what data is allowed, require verification and name an owner.
- Train. Show real examples of good prompts and bad ones.
- Review regularly. Tools and guidance change quickly.
Questions clients may ask
Corporate clients increasingly ask in outside counsel guidelines and questionnaires whether and how the firm uses AI. Having a clear, truthful answer ready is better than improvising. Some clients prohibit AI use on their matters entirely, so check agreements before using any tool.
What to watch
Court rules on AI disclosure vary by judge and jurisdiction, and standing orders change. Assign someone to track those in the courts where you practice. Also keep an eye on how vendors change terms over time.
Closing thought
AI is neither a miracle nor a malpractice trap. It is a tool, and the risk depends on how it is selected, configured and supervised. Counsel Cyber helps law firms evaluate AI vendors, set up sanctioned tools and write policies that match real workflows. If you want a practical starting point, we are glad to talk it through.