You’re shipping an AI product in the United States. Data provenance, the claims you make and the contracts you sign this quarter set your exposure.
Information, not advice
This page is information, not legal advice about your matter. A retainer begins only when a lawyer confirms it.
Start here
Write down what your model is trained on, and what you say it can do
Those two facts sit behind every enterprise contract you sign and every claim the FTC might test. They are cheap to pin down now and expensive to reconstruct later.
What to work through, in order
Data provenance
Fair use in training is actively litigated with no settled answer. State privacy laws, notably the CCPA as amended, create separate obligations around personal information and automated decision-making.
Claims about the system
The FTC has repeatedly warned against unsubstantiated AI claims and has brought enforcement actions. Substantiation has to exist before the claim is made, not after it is challenged.
Contracts and liability
Indemnities for IP claims arising from training data have become a standard enterprise ask. Limitation of liability caps are usually the most heavily negotiated term.
Next steps
What happens if you go ahead
Scoping call
Written scope and fixed-fee quote
Engagement, once a lawyer confirms
Who would handle it
Offices
New York
450 Park Avenue, 3rd Floor, New York NY 10022
San Francisco
28 Geary Street, Suite 625 #133, San Francisco CA 94108