Beware, Deepfaked AI Photos Might be Coming for You
Recently, an artist who created remarkably realistic AI deepfakes photos of politicians cheating on their spouses was banned from Midjourney.
Los Angeles-based satirist and video editor Justin T. Brown used the widely popular AI image-rendering platform Midjourney to create his unusual project.
With the Midjourney AI, he was able to create a whole series of fake photos of powerful politicians like Donald Trump, Joe Biden and Barack Obama.
The impressively realistic images took about an hour each to create according to Brown, but they also got him banned from the Midjourney platform.
Brown claimed to have made the visual renderings as a form of social commentary. To this end, he titled the entire series “AI will revolutionize the blackmail series.”
He then claimed that his hope was to underscore how AI technology could become a weapon for hurting people arbitrarily.
As the artist explained to the website Petapixel, “I created the series with the goal of dragging some of the most influential people in America into the conversation about the dangers of AI, and reach the biggest possible audience,”
He added that his work aimed to “create buzz and spark conversation in the most salacious, shocking and funny way possible.”
He also explained that the project itself had to be threatening for the sake of underscoring the reality of the threat with visual AI technology.
Considering how realistic some of these images look, and that he ended up being banned from Midjourney, it’s hard not to agree that he struck a spark.
To his credit, Brown never tried to misrepresent the “photos” as real, but for Midjourney, -a site obviously averse to creating bad PR with its AI technology- that wasn’t enough.
Within just a day or two after posting the visuals on social media and on the Midjourney subreddit of Reddit, he found himself banned by the platform.
He specified that he wasn’t just blocked from using certain specific prompts, instead, he’d been completely blocked from using his account.
In typical tech-corp fashion, Midjourney wouldn’t tell him why he was banned, but the timing and his project make the reason blindingly obvious.
If anything though, Midjourney’s reaction made things worse for the platform.
Brown’s images themselves had already gone widely viral and garnered over a million Twitter views and the Midjourney ban only added media fuel to the social sharing fire.
On the other hand, while this banning of the artist is interesting, it’s also rather secondary to the broader implications of his work:
In the case of Brown’s own images, the very famous figures involved and the extreme improbability of each “shot” happening in real life makes it really easy for most people to disbelieve them right off the bat, especially in an age of photorealistic AI rendering.
However, if someone else with much more malicious intent were to create somewhat more subtle AI renderings of less famous people, or of personal enemies in ordinary life, the context changes completely.
This applies especially in today’s social media-connected world. Because of it, a single public posting of a very realistic fake image of someone to that person’s social or work circles could rapidly stain their reputation, or even ruin their life.
Brown himself stated, “This kind of disinformation will continue to happen, next time by bad guys with bad intent, so we better start getting a handle on it now.” He’s absolutely right.
The cat’s already out of the bag on AI deepfakes. They’re here to stay and they’re already being used for much cruder purposes than social commentary or artistic expression.
Midjourney simply banning such things without participating in a serious conversation about fighting their uglier possibilities is a step in the wrong direction.
For that matter, so too is the company’s blatant pandering to very specific political interests on other occasions.
Instead, the platform could have used a stunt like Brown’s as a perfect opportunity to discuss how it can do its part to counter the genuinely destructive use of deepfakes while also preserving their legitimately creative possibilities, even if these are politically edgy.
Since that didn’t happen, other cases of this will keep popping up regardless of any specific account bans, and a needed debate will continue to be postponed.