Our company is built around a fundamental guiding principle: We put creators first. In a complicated digital world, we advocate for a future where creators play the most important and valuable role in online media.
When it comes to ChatGPT, AI, and related emerging technological developments, we’re leveraging the scale of our community and making moves to protect creators’ interests.
Generative AI is rapidly expanding and evolving, with new capabilities and concerns emerging every week. Right now, AI companies are running the show, seeing huge benefits and making speedy advances by training their models on content others have created.
Behind the scenes, we’ve been working quickly with others around our industry to help our creators stay ahead of the curve.
Creators have the right to decide how their content is used
We’ve heard many creators say they feel like generative AI systems have co-opted their original work to train the models that generate content. This AI-generated output might then be used to compete with you or to help conversational search engines reduce the likelihood of users clicking through to your site.
We hear your concerns.
We believe it’s your fundamental right to decide how others use your content and on what terms, including the right to determine when AI companies can train models based on your content.
If you’re among those who feel these companies are building tools to profit from your creative efforts without permission, compensation, or even acknowledgment, we want to give you more control.
Today, we’re happy to introduce a first specific step you can take to broadcast if you don’t want to allow your content to be used to train AI tools without your consent.
We’re sharing sample Terms of Service language you can add to the legal notices on your site. This language won’t immediately stop generative AI systems from training on your content. But it allows you to draw a line in the sand and state that it’s explicitly against the terms under which you make your site’s content available.
To complement these terms, we’re also testing out capabilities that can add a machine-readable signal to your pages to indicate you don’t want the content used for purposes of training AI models.
Step 1: Sample Terms of Service language
This language creates the argument that any user (including a person or company that directs a bot) that accesses your site for purposes of training generative AI is violating your terms, and therefore breaching the contract the user enters into when using the site.
The owner of this website does not consent to the content on this website being used or downloaded by any third parties for the purposes of developing, training or operating artificial intelligence or other machine learning systems (“Artificial Intelligence Purposes”), except as authorized by the owner in writing (including written electronic communication). Absent such consent, users of this website, including any third parties accessing the website through automated systems, are prohibited from using any of the content on the website for Artificial Intelligence Purposes. Users or automated systems that fail to respect these choices will be considered to have breached [these Terms of Service][this Agreement].
(Replace “[these Terms of Service][this Agreement]” with the title of your site’s Terms of Service or similar page that contains this material. If you created a standalone “No AI” page, omit that last sentence.)
Note that CafeMedia/AdThrive cannot enforce this contract between you and the user. Any enforcement efforts would be up to you. But with this language, you can establish your policy and add your voice to the growing movement for creators’ rights when it comes to generative AI.
Looking ahead: machine-readable “No AI” Signals
Updating your site’s terms of service is just the beginning.
The industry also needs to define and adopt a standard, machine-readable convention to indicate content should not be used for training generative AI. This could build on the existing use of robots meta tags (which in practice are often managed by SEO plugins like Yoast).
We’re fond of the approach pioneered by DeviantArt, which involves inserting a new type of robots meta tag like this into page headers:
<meta name=”robots” content=”noai, noimageai”>
But this approach entails some challenges. The “noai” and “noimageai” values for the robots meta tag aren’t industry-standard yet, and we don’t have assurances that any or all bots will respect it. What’s more, we’ve found certain SEO plugins rewrite robots meta tags on their own, stripping these tags from the page.
We’re continuing to test approaches like this and collaborate with partners across the industry to pursue wide adoption of a standard and continue to address this concern from multiple angles. Stay tuned.
We’re excited to offer something you as a creator can do now to assert your rights right now.
Still, the questions here remain numerous. The US Copyright Office, for example, recently noted that training generative AI on copyrighted content raises issues that need investigation (see the end of Section II in this publication.) We expect it will be some time before anything official comes out of this, but consider it a good sign that this issue is receiving broader attention.
The bottom line is that we’re committed to navigating the evolution of this technology with you and working to ensure creators are protected and represented every step of the way.