Midjourney, a groundbreaking artificial intelligence image generation company, is using its software to avoid being used to distract from or misinform the 2024 U.S. presidential election. It is considering banning the creation of political images of Joe Biden and Donald Trump.
Midjourney CEO David Holtz said last week that “we don’t know how much we care about political speech in the coming year on our platform,” adding that the company does not support political images such as It added that it was at the stage of “hitting”, or banning. Potential presidential candidates “for the next 12 months.”
In a conversation with Midjourney users in a Discord chat room, Holtz went on to say, as reported by Bloomberg: Trump is really interesting aesthetically. But perhaps it’s better to retreat a little during this election than to do nothing at all. Let’s take a look. ”
AI-generated images have been of recent concern. Two weeks ago, a pornographic image featuring Taylor Swift’s likeness prompted lawmakers and Swiftie supporters to demand stronger protections for AI-generated images.
Swift’s image was posted on 4chan, a community bulletin board that is often linked to the sharing of sexual, racist, conspiratorial, violent, or other anti-social material, with or without the use of AI. Let’s go back.
Holtz’s comments come as safeguards created by image generators play a cat-and-mouse game with users to prevent them from creating questionable content.
AI in politics is a growing concern, but the MIT Technology Review recently noted that discussions about how AI threatens democracy “lack imagination.”
“People are talking about the dangers of campaigns that attack adversaries with fake images (or even fake audio or video), because we already have decades of experience dealing with doctored images. ,” the review notes. He added that it is “hard to attribute the surprising election results to any specific AI intervention.”
Still, image generation company Inflexion AI announced in October that its chatbot Pi will not be allowed to advocate for any political candidate. Co-founder Mustafa Suleiman said at a Wall Street Journal conference that chatbots “maybe [have] Even if it works perfectly, it remains a human part of the process.
Meta’s Facebook announced last week that it plans to label posts made using AI tools as part of a broader effort to combat election-year misinformation. Microsoft’s OpenAI has announced that it will add watermarks to images created on its platform to combat AI-generated political deepfakes.
“Protecting election integrity requires cooperation from all corners of the democratic process, and we want to ensure that our technology is not used in any way that could undermine that process,” the company said in a blog post last month. said.
“My biggest fear is that new capabilities in AI will lead to better deepfakes than we had in 2020,” OpenAI CEO Sam Altman said at a recent event. he said.
In January, a fake voice call impersonating Joe Biden telling voters in New Hampshire to stay home highlighted the potential for AI to manipulate politics. The FCC later announced a ban on AI-generated voices in robocalls.
“What we’re really seeing is the rapid increase in innovation and how society is working to understand best practices, codes of conduct, what we should be doing, what new laws should be. There’s a disconnect between our ability to come together and our considerations’ – and it’s still painfully slow,” said David Ryan Polgar, president of the nonprofit All Tech Is Human. He previously told the Guardian.
Midjourney Software was involved in creating a fake image of President Trump being handcuffed by agents. Other images posted online include images of Mr. Biden and Mr. Trump working together to knit a sweater, Mr. Biden grinning while firing a machine gun, and Mr. Trump meeting with Pope Francis at the White House. Published.
The software already has many safety measures in place. Midjourney’s Community Standards guidelines prohibit images that “depict public figures or events in a manner that is disrespectful, harmful, or misleading, or that is likely to mislead.”
Bloomberg noted that what is allowed and disallowed depends on the version of the software used. The older version of Midjourney produced an image of President Trump covered in spaghetti, but the newer version did not.
But even if Midjourney bans AI from generating political images, consumers, including voters, likely won’t notice.
“They’ll probably just bash it and say nothing,” Holtz said.
