Midjourney, a popular artificial intelligence imaging platform, has decided to suspend its free trials after a series of scandals involving some of its creations. These include those that have gone viral in recent days with the false arrest of Donald Trump and the viral images of Pope Francis in a Balenciaga-style coat.
According to David Holz, CEO and founder of the platform, the measure was taken due to the “extraordinary demand” and to “abuse” by the users.
Viral image scandal prompts Midjourney to revise its content standards and suspend its free service
Prior to the suspension, Midjourney allowed anyone who signed up via Discord to generate up to 25 free images. Only after this quota was exhausted, did he charge subscriptions that range between $10 and $60 per month. The platform had earned the loyalty of more than 13 million members in just one year since its launch in 2022.
In mid-March of this year, Midjourney released version 5 of its service, which significantly improved the quality of images depicting people. However, shortly thereafter, the generated images of Donald Trump and Pope Francis went viral, exposing the platform’s ability to fabricate quite convincing fake news. Holz said they were reviewing the site’s moderation guidelines, especially those having to do with generating content that includes real people. “The images become more and more realistic and the tools more and more powerful”he said in a live audience with some 2,000 people that was held today on Discord, as reported by the Washington Post.
Midjourney’s content standards are more permissive than those of DALL-E, the OpenAI image generator, but more restrictive than those of Stable Diffusion, from the Stability AI company. Users must be at least 13 years of age to use Midjourney. The Terms of Use prohibit adult content and gore, as well as text indications that are “inherently disrespectful, aggressive or abusive”. However, the content standards are not very restrictive. The journalist Eliot Higgins, responsible for the images of Trump, was expelled from the platform after the material went viral, but the company did not explain the reason.
Midjourney was sued last January by an artists’ collective, alleging that the developers trained the AI model on images and works posted online without the consent of the original creators. The company has not yet responded to the claim in court.
Dall-E, by contrast, prohibits the creation of images that include politicians, protests, or other content that could be used for campaigning. Also those that intend to refer to “major conspiracies or events related to major ongoing geopolitical events”.
Although deepfakes can be used for friendlier and more practical purposes, cases like this show that AI needs certain safeguards that guarantee safe spaces.
