For lawyers

Legal reasoning is dense, cited, and rarely obvious to non-lawyers. That is exactly why frontier AI labs have quietly become one of the largest new employers of practicing lawyers on an hourly basis. Contract analysis, statutory interpretation, procedural reasoning, and adversarial argumentation are all fields where the difference between a plausible answer and a right one costs someone something real.

The rate at which models handle basic legal prompts has improved dramatically. The rate at which they handle the specifics of contract clauses, jurisdictional nuance, or the difference between what a statute says and what a court has actually done with it has improved much more slowly. That gap is where the work is.

What the work looks like

• Response evaluation. You review model outputs on legal prompts (contract clauses, court reasoning, statutory interpretation, motion drafting) and rank them, correct errors, and flag anything unsafe or unauthorized.
• Question authoring. You write high-quality legal vignettes at the level of a bar-exam question or a client memo, dense enough for models to learn from.
• Rubric review. You audit the criteria being used to score legal outputs. Rubrics that pass in a technical group often do not survive a lawyer's read.
• Adversarial testing. You elicit responses where the model is confidently wrong in a way another attorney would catch immediately.

The work is asynchronous. Blocks of two to four hours. Structured deliverables. No court appearances, no client management, no billing hours.

What labs actually look for

Active bar admission, or verifiable active practice in a jurisdiction. Recent work in litigation, contract, regulatory, corporate, tax, or another substantive area. Written English at the level any peer-reviewed law journal expects.

They do not look for AI experience. Almost no lawyer hired for this work has ever trained a model. They are hired for their judgment on legal correctness, not their technical skill.

Where the pay lands

Rates are set by the labs and are in USD. In most jurisdictions, hourly compensation for this work compares favorably to per-hour billing rates in local practice, with none of the overhead. That gap is largest in emerging markets, where the same expertise is priced very differently by geography.

How to enter

Sign up, upload a recent CV that shows your practice area clearly, and let the platform surface the roles currently open in legal annotation, legal evaluation, and legal adversarial testing. Apply where the fit is honest. Skip the ones that stretch too far. There will be more.