Attorney-client privilege extends to the technology attorneys use to communicate and store client information. Sending client documents to a third-party AI service raises questions about inadvertent disclosure and waiver of privilege that courts are only beginning to answer. Running a model locally, where client data never leaves your hardware, is the most defensible posture while the law catches up to the technology.
The legal profession’s hesitation around AI tools is not technophobia, it’s a specific concern about privilege. Attorney-client privilege protects confidential communications between attorneys and clients. The question courts and bar associations are wrestling with: does routing that information through a third-party AI service constitute disclosure that waives privilege?
That answer is not settled. Different jurisdictions are reaching different conclusions. Several bar associations have issued guidance ranging from permissive to cautious. Many malpractice insurers are asking pointed questions about AI usage.
A local model sidesteps most of these questions. Client documents that never leave your machine can’t be exposed by a vendor breach, can’t appear in a third party’s training data, and don’t require you to analyze someone else’s privacy policy.
My legal experiments start with high-volume, format-intensive tasks: contract drafting, research memo scaffolding, and discovery review. These are tasks where the model handles structure and volume, and the attorney handles judgment and certification. The first one is already published: a contract first-draft run, showing the model used and the hardware it ran on.
The key test is not whether the model can replace an attorney, it cannot. What matters is whether it can eliminate the formatting and organizing that eats the non-billable hours between leaving the office and actually getting paid.
Every bar association is different. The ABA’s Model Rules provide a baseline, but state-level variations are significant. Ethics opinions on AI use are being issued and revised regularly. If you’re considering local AI tools in your practice, get a written ethics opinion from your state bar’s ethics hotline before deployment.
Nothing on this site constitutes legal advice. See the Scope & Disclaimers page.
A local Llama 3.1 13B model reviewed five commercial vendor contracts and flagged clauses for attorney attention. It caught most obvious red flags but missed nuanced jurisdiction-specific risks, making it useful as a first-pass triage tool, not a replacement for legal review.
The experiments here cover what's technically possible with local hardware. Your specific regulatory obligations, your state's rules, your specialty's requirements, your malpractice carrier's stance, your EHR vendor's terms, are questions this site cannot answer.
See the Scope & Disclaimers page.