- Item 1
- Item 2
- Item 3
- Item 4
Midjourney Bans Journalist Whose Fake Trump Photos Went Viral
Fake images of Trump getting arrested, created via AI art platform Midjourney, quickly went viral and show how our society may not be ready for the consequences of AI-generated images.
Donald Trump is arrested by the police. Generated via Midjourney by Eliot Higgins. Source: Eliot Higgins
🧠Stay Ahead of the Curve
Eliot Higgins, the founder of open-source investigative website Bellingcat, generated dozens of images of Donald Trump getting arrested via Midjourney and then shared them on Twitter
Despite making clear the photos were fakes, Higgins was banned from Midjourney after the photos went viral, racking up millions of impressions as they rapidly circulated without context on social media
While AI-generated images aren’t perfect, their disruptive potential is already evident in their power to spread misinformation online or stoke outrage
March 22, 2023
It was the experiment that went too far. When Eliot Higgins, founder of open-source investigative website Bellingcat, read about the expected indictment of Donald Trump, he had an idea. Why not visualize the arrest by generating images of this event through the AI art platform Midjourney?
Even though Higgins, 44, is a seasoned journalist who deeply understands the power of media, he never anticipated the impact his images would have. Within two days of sharing these images, they had racked up over 5 million views on Twitter and had seen countless reshares on other social media platforms as well.Â
— Eliot Higgins (@EliotHiggins) March 20, 2023
Twitter put their own label on these images. “Missing Context. Independent fact-checkers say information in this post could mislead people,” read one warning. Yet the warning came too late. By then, these images had circulated far and wide. Numerous publications, including the conspiracy site Infowars, had featured these images – many with their own political spin.
More concerning were users on Twitter who believed the images were real, and social media users who then took the images and reshared in an attempt to spread deliberate misinformation. As the Washington Post reported, one Facebook user shared the photos to their 9,000-plus followers with the caption:
“In an unprecedented turn of events, former President Donald Trump was arrested and escorted to federal prison. The shocking image of Trump, with officers holding both of his hands, quickly circulated on social media.”
As the images continued to circulate, Midjourney responded with more than just a warning; they banned Higgins from using their platform. “I suspect it was pushing my luck when I did the [Twitter] thread,” Higgins said. He received no additional communication from the company about why he was no longe able to access Midjourney.
While AI-generated images are far from perfect – six-fingered hands and other strange appendages are common – the current form of the technology is already sufficient enough to generate images that can spread misinformation or undermine the public’s trust in the veracity of images in general.
And the technology continues to evolve rapidly. Midjourney, which launched its AI image platform in July 2022, is already on version 5. Users on Reddit recently raved about the rapid improvements the latest version brought and how quickly the platform’s capabilities had improved. “I can't find a single uncanny valley about this anymore, it just straight up looks like a photo,” said Reddit user mazi710,
Research
In Largest-Ever Turing Test, 1.5 Million Humans Guess Little Better Than ChanceJune 09, 2023
News
Leaked Google Memo Claiming “We Have No Moat, and Neither Does OpenAI” Shakes the AI WorldMay 05, 2023
Research
GPT AI Enables Scientists to Passively Decode Thoughts in Groundbreaking StudyMay 01, 2023
Research
GPT-4 Outperforms Elite Crowdworkers, Saving Researchers $500,000 and 20,000 hoursApril 11, 2023
Research
Generative Agents: Stanford's Groundbreaking AI Study Simulates Authentic Human BehaviorApril 10, 2023
Culture
As Online Users Increasingly Jailbreak ChatGPT in Creative Ways, Risks Abound for OpenAIMarch 27, 2023