AI training

How to opt your art out of AI training

Updated

Opting out now happens on the site that shows your art. The site tells AI crawlers what they may use it for.

Have I Been Trained is down

As of November 2026, Spawning's Have I Been Trained shows only a maintenance page. You can't search it or add new opt-outs. Opt-outs already in the Do Not Train registry still exist. The page asks AI trainers to contact Spawning to use it.

Last checked November 14, 2026.

On your own website: robots.txt

Add these lines to the robots.txt file at the root of your site. Your site builder needs to let you edit that file.

User-agent: *
Content-Signal: search=yes, ai-input=yes, ai-train=no

User-agent: GPTBot
User-agent: ClaudeBot
User-agent: Google-Extended
User-agent: Applebot-Extended
User-agent: CCBot
User-agent: Meta-ExternalAgent
User-agent: Bytespider
Disallow: /
  • The Content-Signal line says no to AI training and yes to search and AI answers. Content Signals
  • Each User-agent line names a crawler or a training setting.
  • The full list changes often. ai.robots.txt keeps it current.

EU law: TDMRep

EU copyright law lets you reserve your work from text and data mining in a machine-readable way. TDMRep is one way to do it. Serve this file at /.well-known/tdmrep.json:

[{ "location": "/", "tdm-reservation": 1 }]

Or send a tdm-reservation: 1 header with each page.

On sites you post to

  • DeviantArt marks posts noai by default.
  • Meta's form to object to AI training is offered in the EU and UK only.
  • Many sites have no AI setting. Check the ones you post to.

On the image itself: Glaze

Glaze, from the University of Chicago, changes your image file before you post it to make style copying harder. The Glaze app is free. WebGlaze runs it in the browser and is invite-only with daily limits.

Researchers have published ways to strip it. It only covers the files you run through it.

What opt-outs can't do

They don't remove your art from models that are already trained.

They don't stop a crawler that ignores them.

On SievePage

Every SievePage page sends Content Signals and TDMRep and turns away known AI training crawlers. There is nothing to turn on. Search engines and link previews still work. Content protection.

Common questions

Is Have I Been Trained still working?

No. As of November 2026 it shows only a maintenance page. spawning.ai showed one by August 19, 2025, and haveibeentrained.com by August 23, 2025. You can't search it or add new opt-outs. Opt-outs already in the Do Not Train registry still exist, and the page asks AI trainers to contact Spawning to use it.

How do I opt out now that Have I Been Trained is offline?

On the sites that show your art, through robots.txt, Content Signals and TDMRep. Change AI settings on the sites you post to where they exist.

Does opting out remove my art from AI models that are already trained?

No. Opt-outs cover future collection. They don't change a model that was already trained.

Does robots.txt stop AI training?

It asks crawlers to stay out. OpenAI, Anthropic, Google, Apple and Common Crawl say their crawlers follow it. It doesn't stop a crawler that ignores it.

What are Content Signals?

A robots.txt line Cloudflare proposed in September 2025. It states whether you allow search, AI answers and AI training. Crawlers don't have to follow it.

What is TDMRep?

A W3C Community Group spec for reserving your work from text and data mining in a machine-readable way. A site sends it as a tdm-reservation header, a meta tag or a /.well-known/tdmrep.json file.

Does SievePage opt my page out?

Yes. Every SievePage page sends Content Signals and TDMRep and turns away known AI training crawlers. There is nothing to turn on.