AI image generators like Dall-E are the hype this summer and never cease to amaze. But the first internet platforms are slamming the door.
More and more internet portals for art prohibit the posting of images created by an AI image generator. The US blogger Andy Baio has compiled this and at the same time found that two of the most popular platforms, DeviantArt and ArtStation, have not yet reacted – although the calls for banning them from their community are getting louder. In a particularly well-commented post in the DeviantArt forum, the portal is asked to do something. The site is already being flooded with AI images and, as a countermeasure, an upper limit for posts per day is necessary, it says.
“Zero tolerance” for AI art
The platforms Newgrounds, Inkblot and Fur Affinity have explicitly forbidden images from AI image generators. The statements sometimes refer to the fact that the AI ​​images are based on countless works by artists with whom the software was trained. But one wants to support the artists and their content, writes Fur Affinity. AI content is not in the best interest of the community. Newgrounds is similar, but allows for backgrounds that come from an AI – if they are explicitly referenced. inkblot speaks on Twitter of a zero-tolerance policy towards AI works.
Recommended Editorial Content
With your consent, an external video (Kaltura Inc.) will be loaded here.
Always load videos
I consent to external content being displayed to me. This allows personal data to be transmitted to third-party platforms (Kaltura Inc.). Read more about our privacy policy.
Image generators such as Midjourney, Dall-E or, more recently, Stable Diffusion create images based on text specifications. In the past few months, a real hype has developed around such software. The results vary in quality, but Midjourney and Stable Diffusion in particular have been causing astonishment on the web for weeks with the results they have delivered. The associated discussion about art has long since reached the masses. Because the generators can also imitate photographs in a deceptively real way, there are also indications that disinformation has once again become much easier.
Just a few days ago, an image generated by an AI won an art competition in the USA, triggering a global debate. There was some talk of the death of art. Further arguments are now being exchanged in the discussion about AI images on art platforms. The greatest concern is that the image generators can be used to create artistic images in such large numbers and at such a high frequency that hand-made works are lost in the tide. This is one of the main criticisms in the debate on DeviantArt. Others point out that the works come from the AI ​​used, not from the uploader himself.
Open copyright issues
Another point of criticism of AI art is likely to become increasingly important in the coming weeks and months in view of the current hype: Even if it is usually not clear on which training material the algorithms and their works are based, it is becoming clear again and again that many of them are Are works whose creators have probably never consented to such use. For example, Baio has gathered that millions of images from the training material for Stable Diffusion come from Pinterest, where images are collected from other sources. A broad copyright discussion about Midjourney & Co. seems inevitable, also because the AI ​​technology can generate images in the style of certain artists thanks to the training material.
(mho)
