What the current cases are about
Most AI copyright litigation in the US centers on two questions. The first is whether copying works to train a model is infringement or fair use, and rulings so far, including an appellate decision, have differed depending on how the material was obtained and whether the use competes in the original's market. The second is whether specific outputs reproduce protected expression closely enough to infringe. Other claims involve removed copyright management information, breached licensing terms, or datasets assembled from pirated sources. Courts and the Copyright Office have so far required human authorship for copyright protection, which affects who can claim rights in AI-assisted material. Appeals are pending and new decisions keep arriving, so any analysis needs to be checked against the current state of the law.
Records on either side
If you are a rights holder, document what you created, when, and whether it is registered, since registration affects whether you can sue over a US work and which remedies may be available. Collect examples of outputs that resemble your work, with the prompts and dates used to produce them, and keep copies of the terms under which your work was posted online. If you are a developer or a business deploying a model, the important records are data sourcing documents, licenses, vendor contracts, opt-out and filtering practices, and any indemnities your model provider offers. Do not delete datasets, logs, or training records once a dispute is foreseeable, because preservation duties apply here as in any litigation.
Deciding how to proceed
Many matters in this area start as licensing discussions rather than lawsuits, and some never become anything more. Class actions have been filed against several large developers, so an individual creator may already fall within a proposed or settled class, which changes the calculus for a separate case. Businesses using third-party models often have contractual protection worth reviewing before responding to a claim. In an initial meeting we sort out where you stand, which rules are settled and which are not, and what a realistic objective looks like. Given how quickly the field shifts, we also plan to revisit the approach as new decisions come down.
