Your Photos and AI: What Photographers Should Know About Training Data

General information, not legal advice, and this area is changing quickly. For your specific situation, talk to a qualified lawyer in your country.

AI image models are trained on enormous sets of images scraped from the web, and a lot of photographers' work is in there. This is the newest and fastest-moving corner of photo rights, so here is an honest snapshot of where things stand.

What is actually happening

Companies gather huge datasets of online images to train models that then generate new pictures. Two separate questions come out of this: was it allowed to use your photos to train the model (the input side), and is an AI-generated image that looks like or derives from your work a problem (the output side). The law is still being worked out on both.

In the EU: you can reserve your rights, in theory

EU law has a text and data mining exception that lets companies mine lawfully accessible content, but rights holders can opt out by reserving their rights in a machine-readable way, and the EU AI Act has confirmed this applies to general-purpose AI. In March 2026 the European Parliament reinforced this direction. The catch, honestly, is that it is still unclear exactly how an individual photographer effectively signals and enforces that opt-out, and most small creators do not have the resources to police it.

In the US: it is being fought out in court

There is no EU-style opt-out. Instead it comes down to fair use, and the cases are mixed and still running. Some early rulings treated AI training as transformative, while keeping pirated source copies off-limits, and the big image-focused cases are still pending. The Supreme Court has, separately, left in place the rule that works generated purely by AI cannot be copyrighted, because copyright needs a human author. In short: unsettled, and worth watching.

What you can do today

• Keep your copyright and metadata intact on everything you publish (the foundation in copyright basics).
• Where platforms offer an opt-out or a do-not-train signal, use it.
• Know where your work already appears, because the same monitoring that catches a stolen photo on a blog is the starting point for spotting AI-related reuse.

Where we are headed

Catching AI-generated images that are derived from or trained on a photographer's work is the next frontier, and it is exactly the problem we want to help solve. For now, the foundation is the same as it has always been: own your rights, keep your evidence, and know where your work is.

Start with the basics: know where your photos are being used. Start a free account and let Pixel keep watch.