Midjourney temporarily disable free trials because of abuse ⛔

Cliquez ici pour lire en français

Midjourney is a victim of its own success. The AI-based image generation platform is becoming increasingly popular, which comes with its own set of problems. And that’s what pushed its developers to end the free trial offer.

Abuse is on the rise on Midjourney⛔

Midjourney is an AI for generating images from text. We presented it in a previous article, which you can find by clicking here. In that article, we presented some images generated by the AI that had created a huge buzz on the internet, like this photo of the pope, which is completely false.

Source : Reddit

Well, all this buzz made new users discover the platform, who then rushed to generate images.

« Due to a combination of extraordinary demand and trial abuse we are temporarily disabling free trials until we have our next improvements to the system deployed » said David Holz, Midjourney’s CEO, in a post on Discord.

As you can read in the statement of the CEO of Midjourney, this new influx of users has not been without consequences. In particular, he deplores many misuses, with representations of real people. And since « the images are more and more realistic and the tool is becoming more and more powerful », this facilitates the recrudescence of fake news.

Initially, Midjourney was made available again for free with new security measures. Unfortunately, they turned out to be insufficient, and monthly subscription is now required. Despite this, the Midjourney teams will continue to improve their system, as the CEO confirmed to The Verge.

Moderation is hard and we’ll be shipping improved systems soon. We’re taking lots of feedback and ideas from experts and the community and are trying to be really thoughtful », a déclaré Holz à The Verge.

The Xi Jinping case raises questions🧐

Among the plethora of generated images featuring heads of state and other political figures, few involve « Xi Jinping » or the « Chinese president ». And the reason is simple: Midjourney simply prevents the creation of images of the Chinese leader. To defend himself, the CEO David Holz explains that it is « To avoid drama » because « political satire in China is rather unpopular ». This statement, which is anything but innocuous, raises an even more important question: Who sets the rules? If the Chinese president can be protected from hijacked images, why not others?

Midjourney keeps a list of banned words « related to topics in different countries, based on complaints from users in those countries, » according to a post by David Holz from last October. But it is not sharing the full version of this list in order to minimize « drama. » In the absence of uniform regulations set by regulators, companies, including Midjourney, are responsible for determining what is and is not allowed.

Midjourney, Stable Diffusion and Dall-E all follow different rules. For example, Dall-E prohibits the use of images of political figures or photos related to major ongoing geopolitical events, which is quite broad. Stable Diffusion has few rules in place. Midjourney falls somewhere in between, banning gory images and text messages considered « inherently disrespectful, aggressive or otherwise abusive. »

Sources : The Verge, O1net