The solo attorney's guide to safe legal AI
You're the attorney, the paralegal, and the person who catches the mistakes. Here's how to get leverage from AI without betting your bar card on it.
For a solo, AI is the most tempting and the most dangerous tool available. Tempting because you have no one to delegate to. Dangerous because you also have no one to catch the error before it's filed under your name.
Rule one: never cite what you can't open
Treat any citation you didn't personally read as unverified until you've opened the source. A tool that makes this easy, by linking every cite to its source, is doing real work. A tool that just hands you text is handing you risk.
Rule two: keep the human checkpoint
Use AI to remove the drudgery (the first draft, the chronology, the initial research pass), not to remove your judgment. The moment you're accepting output you haven't reviewed, the leverage has become a liability.
Rule three: mind confidentiality
Your ethical duty to protect client information doesn't pause for a convenient tool. Understand where your documents go, whether they're used to train a model, and what the provider is contractually bound to. "Your data never trains a model" should be in writing, not just on a webpage.
Rule four: choose verification-first
The safest legal AI is the kind built to prove its answers rather than assert them. For a solo with no safety net, that property isn't a nice-to-have. It's the whole point.