How to Protect Your Photos from AI: A Practical Guide for Photographers

The world of photography is changing rapidly. As AI models scrape the web for data, many photographers are asking: How can I fight back in this new digital landscape? We have outlined 6 practical steps to make a robot's job harder and a look at how the law currently views AI-generated content.

1. Block the Bots (Robots.txt)

AI platforms crawl websites using "crawlers" or bots. You can give them a clear "No Entry" sign using a robots.txt file on your website’s backend. This tells search engines and AI bots which folders containing your photos should not be indexed.

• Robots.txt: Disallow crawlers from entering your image directories.
• Low Resolution: Upload images to social media at a quality that is "eye-pleasing" but insufficient for training high-end AI models.
• WordPress Plugins: Tools like Kudurru can detect AI bots and block their IP addresses or serve them corrupted images.

2. Utilize New Metadata (IPTC Standard)

The IPTC organization recently updated its metadata standards. You can now embed a "Data Mining" field directly into your photo's metadata, stating that you do not wish for the image to be used for AI training. This information stays with the file even when shared.

How to do it in 4 steps:

1. Use a tool that supports IPTC Photo Metadata (e.g., Adobe Lightroom, Bridge, or ExifTool).
2. Open the Metadata Viewer.
3. Find the XMP property plus:DataMining.
4. Set the value to DMI-PROHIBITED-GENAIMLTRAINING (Strict prohibition for Generative AI).

Pro Tip: While protecting your assets from external bots is crucial, internal organization is equally important. Read our guide on Studio Management and Workflow: The Key to a Successful Photography Business to ensure your protected files are archived and managed efficiently.

3. Adjust Your Social Media Habits

• Watermarks: While AI can remove them, they remain a clear legal signal: "This photo has an owner."
• Privacy Settings: Check your settings on Facebook, Instagram, or TikTok. Many now have (often hidden) "opt-out" options to prevent your data from training their proprietary AI models.

4. Deploy "Digital Poison" (Glaze and Nightshade)

Researchers at the University of Chicago have developed tools that "cloak" or "poison" data to protect artists:

• Glaze: Applies an invisible layer that makes AI see a different artistic style than humans do, preventing style theft.
• Nightshade: "Poisons" the data. If enough artists use it, an AI model scraping these images will lose its ability to distinguish basic objects (e.g., seeing a dog as a cat).

5. Follow the Legal Landmarks

The protection of creators is currently being decided in courtrooms. For example, Getty Images is suing the creators of Stable Diffusion for using 12 million of their images without permission. The outcome of these cases will determine the "Fair Use" rules for everyone in the industry.

6. The Legal Perspective: Who is the Author?

In most jurisdictions, including the Czech Republic and the EU, authorship is tied to human creativity. Without a "human creative contribution," AI outputs are generally considered "public domain."

• USA: The US Copyright Office has rejected registrations for AI images (e.g., Steven Thaler’s works) due to a lack of human authorship.
• UK: Takes a more progressive stance, recognizing the person who "makes the arrangements" for the creation of the work.
• Czech Republic & EU: We are still waiting for a unified AI Act. Until then, you must rely heavily on well-drafted contracts.

Final Advice: Read the Terms of Service

If you use AI tools for your own editing, read the TOS. By uploading your photo to a generator, you might be granting that platform a license to use your work for their own training.

Are you worried AI will replace you? We believe the human touch is irreplaceable, but your workflow must adapt.

This content is based on materials from the Fotoprávo and Law4art projects.

‍