Previously, artificial intelligence technology was used to fake photos of Trump in custody. The CEO of Midjourney, David Holz, hinted that the company is on the verge of banning photos similar to those of Biden and Trump.

• Artificial intelligence algorithms were used to create fake photos of Trump in custody.
• The CEO of Midjourney, David Holz, said the company has almost completed the removal of such images from its platform.
• Midjourney applies policies that prohibit the creation of „portraits of events” and „misleading public figures” that can lead to disinformation.
• Midjourney’s decision to exclude such photos will protect the platform from harsh reactions during the election season.
• Other digital companies are developing technologies that use artificial intelligence to recognise and indicate whether a photo, video or audio clip has been created.

Creating fake photos to spread political disinformation ahead of the upcoming US presidential election is easier than ever thanks to the development of artificial intelligence algorithms that can quickly create altered photos and videos. Bloomberg reports that Midjourney’s approach could be to ban political photos altogether. The CEO of Midjourney, David Holz, has reportedly told users in a conversation on Discord that the company is on the verge of banning photos showing Biden and Trump within the next 12 months.

Participants were told by him: „I know it’s fun to take photos of Trump, and Trump has a pretty captivating look. But it’s probably best to just not do it, it’s better to hold back a little bit during this election.” As Bloomberg reports, fakes of Trump’s arrest have been created in the past using the company’s artificial intelligence. Following the viral success of these images and the infamous fakes showing the Pope in a Balenciaga-like coat, the company has completed a free trial period for its AI image generator.

The company now has a policy that prohibits the development of „portraits of events” and „misleading public figures” that „have the potential to mislead”. Using an outdated version of the Midjourney system, Bloomberg was still able to create doctored photos of Trump in spaghetti, but a newer version was unable to create doctored photos of the former president. Of course, even if Midjourney excludes photos of well-known politicians, it’ll only protect its platform from backlash and unwanted attention this election season. It won’t prevent the spread of fake material to influence elections in general or the use of artificial intelligence techniques in political disinformation campaigns.

Other digital companies have also taken steps to prevent or at least more easily recognise political disinformation. While Meta is developing technology that recognises and indicates whether a photo, video or audio clip has been created using artificial intelligence, ChatGPT will soon start labelling photos created with DALL-E 3.