The Human-Made Art Society

The EU Just Gave Artists Real Teeth. Time to Use Them.

The EU AI Act is now enforceable, and for the first time, artists have actual legal tools to protect their work from being scraped into AI training datasets. This isn't a symbolic victory. This is leverage. And if you're not using it, you're leaving protection on the table.

What the Law Actually Says

Starting this year, AI companies operating in EU markets must check whether a data source has a copyright reservation before using it in training. They must publish public summaries of what datasets they used. They must label AI-generated content. And if they don't comply? Penalties up to €10 million or 2% of annual turnover. That's not a slap on the wrist. That's real money.

Under the EU Copyright Directive, you can now reserve your rights and explicitly prevent your work from being used in AI training. Companies are legally required to honor that reservation. The gray area of "we scraped it from the web, so it's fair game" is closing, at least in Europe.

The Problem: They're Still Ignoring You

Here's what's actually happening. Artists are sending opt-out letters. They're emailing AI companies directly. They're doing everything right. And they're getting nothing back. No response. No acknowledgment. Companies are betting that individual artists won't have the resources to enforce their rights.

Sound familiar? It's the same playbook tech platforms have run for years. Move fast, scrape everything, and deal with the legal consequences later, if they come at all. The difference now is that the law is on your side. The question is whether you'll make them feel it.

What You Can Actually Do

First, add a machine-readable copyright reservation to your website. The EU's text and data mining rules require AI developers to check for this before scraping. A robots.txt directive isn't enough on its own, but combined with explicit rights reservations in your terms or metadata, you're building a paper trail.

Second, document everything. If you've sent opt-out requests and been ignored, keep records. When enforcement actions ramp up, and they will, those receipts matter.

Third, join forces. Individual artists get ignored. Organized groups get lawyers. The creative unions, artist collectives, and advocacy organizations pushing for enforcement need numbers behind them. Your voice adds weight.

The EU didn't give artists these tools because it was easy or inevitable. It happened because creators pushed for it. Now the work is making sure those tools actually get used.

Have you tried opting out of AI training? What happened when you did?

5

23 Comments

Sort by:
Linnie AikensMay 18, 2026

@Nick Friend Thanks for this great update! I have some practical questions as an older, less techy person:

1) Is there someone here or at ASF who can provide training on how to add a machine-readable copyright reservation to our websites?

2) When you say "opt-out requests" are these sent if "they" contact us first, or how are we to know where to send these request to be proactive (I'm totally lost on this) And what would such a request say? and

3) Would ASF members who have been verified Human-Made Art be considered a "joint force" under the ASF or AH banner, since we have so many members? If not, is that a possibility as part of that membership? I ask because a number of us artists are solitary and don't have access to a group.

What are your team's thoughts on these questions?

4

Great questions!

0

Questions: 1. where would someone even start to know if there art had been data scrapped? 2. Do ASF websites have machine readable copywrite reservation?

4

Is like to know the same!

0

This is usually an artist's initiative. They embed copyright and tracking into the image itself (using Photoshop, Digimarc, OpenStego, or another tool). Invisible to the eye, steganography cannot be removed like a header. And then using a paid service, track inappropriate usage, scraping, and copying without permission. Digimarc and Steg.AI use specialized crawlers to detect unauthorized copies. Most others require you to find them, but provide the proof you need to file suit.

0

I suggest that the machine readable copyright be a service included with our ASF website subscription.

3
Mary Planding5d ago

Interesting. I just did a search (using Duck Duck Go) and it came back with a whole bunch of YouTube videos and applications that let the user strip meta data including EXIF copyright data from a file.

Article worth reading: https://www.removewatermark.org/blog/is-removing-watermarks-illegal

One major challenge, of course, is not every country adheres to the Berne Convention or this new EU AI Act (which I gather supersedes the EU Copyright Directive of 2019). You can search and discover which countries do / don't follow international intellectual property rights.

2

I was wondering the same thing.... It's an EU law, how does that help us if we are in the States?

1

It never does.

0
Mary Planding6d ago

A really "simple" way to ensure your art has a copyright tag attached to it is to use Photoshop. Underneath "File" there is an "Info" selection. Choose it and it will open up a series of panels. Fill out the information, especially in the sections where it asks for Copyright, etc. Fill it all out with the work's title, your name, your city/state/country (at minimum), fill out rights & permissions (who to contact), etc., etc. When you click "done" immediately SAVE the file. You will notice at the top of the tab for the work a © with the title of your work. When you export the file for use, be sure to choose the option to "Include Copyright info" at minimum (you can choose all meta data or just that). When you click on the exported file, you should be able to "get info" on that file and embedded in that will be the notice of copyright.

I do not know if other software offers this feature. But I've been using it for the past 15 years.

Extra hint: go to https://www.iptc.org/standards/subject-codes. Find the ones that apply to each work and include those. Those are INTERNATIONAL photo industry standard metadata codes for administrative, descriptive, and copyright information about images.

1

I'd have to test to see if it is still true, but Photoshop's "Save for Web" strips metadata from the file to make it "smaller" for faster loading on websites. Simply saving as JPEG should not strip the metadata, the compression of the file only affecting the image data.

1
Mary Planding5d ago

I don't use "Save for Web" so I've never checked that. Good of you to bring that up.

0

I use Affinity apps, I'll have to check if Affinity Photo has the same options. Is think it does. This is good info.

0
TOM ANGMay 18, 2026

@Suzanne Grippi Art check out https://haveibeentrained.com/

but bear in mind the training sets are much much larger than covered by this site.

0

Oooo.... That site is 'under maintenance' at the moment.

0
TOM ANG3d ago

Yes; and seems to have been for some time. The short answer is that if the LLM have not used one of the publicly available databases, I don't think there's any way of knowing if your material has been used. For one thing, the copy they make is small - probably not bigger than 2048 pixels square - and it's temporary.

0

It is your responsibility when you save content (images) to include EXIF data that declares your copyright. Software like Adobe Photoshop does this in the metadata under FILE > FILE INFO. Other software does this too. If you're lazy about saving your images with embedded copyrights, that's on you. ASF cannot add this for you to your images.

0
Kris MercerMay 18, 2026

But some software strips this when it’s uploaded to their website. A lot of marketplaces selling paintings strip all this out.

0

It would have to be deliberately sabotaging images to delete the EXIF information containing copyright in them, which is illegal. Any marketplace found to be doing that would face DMCA fines for every image they did it to.

1
Kris MercerMay 19, 2026

Many marketplaces already strip metadata automatically when images are uploaded and resized. This is standard practice across the web and is generally done to reduce file size and remove unnecessary information such as camera settings and GPS data.

So unless a large number of major websites and marketplaces are already breaking the law, the situation is clearly more nuanced than simply saying that removing metadata is illegal in every case.

My point was that if copyright reservations are embedded in the original file, they may never reach the publicly accessible version because the hosting platform has removed them as part of its normal processing workflow.

That does not remove copyright protection, but it does raise a practical question: if AI companies are required to check for machine-readable copyright reservations, how effective is that safeguard if many platforms strip that information before the images are ever scraped?

0

I'd have to test to see if it is still true, but Photoshop's "Save for Web" strips metadata from the file to make it "smaller" for faster loading on websites. Simply saving as JPEG should not strip the metadata, the compression of the file only affecting the image data.

0

Good to know!

0