A few months ago, DeviantArt got into some controversy over their hosted works being including in AI datasets. In response, they did a blog post.
Their solution to the problem is a one-two punch of imagination, backed by policy.
The imagination bit is, they are adding these meta tags to their heads of their pages, and encouraging developers to do the same.
<meta name="robots" content="noimageai" />
<meta name="robots" content="noai" />
It sure is code. But it means absolutely nothing, and does absolutely nothing. The “noimageai” and “noai” tags do nothing, because DeviantArt just made them up; they are not a part of any official specification.
The only real meaning they have is defined in their policy, where they state:
In order to remain in compliance with DeviantArt’s updated Terms of Service, third parties that continue to use DeviantArt-sourced content to train machine-learning models of any kind must ensure their training data set excludes all content for which either of these directives are present.
Is anyone at all really worried about violating DeviantArt’s terms of service?
Scraping the Internet for data tends to be fairly anonymous, and I sincerely doubt DeviantArt will invest money to follow up on any known infringements with legal action. So, I reckon this is all toothless and performative.
All of this said, I am adding these tags to both Pangolin and Kookaburra in our next update. There is no harm whatsoever including meta tags on pages, nor in politely asking the robots to move along. And if there’s enough groundswell around using them, maybe these tags will actually become recognized by those wanting to ethically source their datasets.