Video creators were quick to capitalize on last night’s Eagles win at the Super Bowl, producing ultra-realistic deepfake clips of the losing team, including Chiefs quarterback Patrick Mahomes.
In the NSFW video below, the AI Mahomes’ voice is so realistic that it’s not immediately clear it’s fake. The video has been viewed over 6.4 million times, with 458K likes as of this writing.
Upon further examination, we can see a slight twitch of AI Mahomes’ mouth as “he” talks, but otherwise the footage seems near perfect. The clearest signal that it’s AI is the fine print at the bottom of the video: “Creator labeled as AI-generated,” which is easy to miss amid a mindless social media scroll.
“This is the most realistic AI I’ve ever seen,” says one commenter. Another adds: “You win the internet for this,” while someone else jokes: “I don’t know, this might be real.”
In Mahomes’ real post-game press conference, the devastated quarterback takes “all the blame” for the loss. The Kansas City Chiefs were gunning for their third consecutive win, something no other NFL team has done before.
The account behind the deepfakes is ScaryAI, which posts parody AI videos of athletes like Steph Curry and LeBron James “talking” about recent controversies and events. After the Golden State Warriors lost to the Los Angeles Lakers in January, ScaryAI posted a video that shows Steph Curry requesting to be traded to the Lakers in a faux press conference.
The account’s bio urges people to get the Parrot AI app “to make videos like mine.” Scary AI does not seem to be owned or operated by Parrot. The AI app has a creator program whereby videos made using Parrot that get over 1 million views earn $100, and the most viral videos of the month get $5,000. Parrot links out to a few examples, and several are from Scary AI.
These videos are legal and are relatively harmless. The viewer may feel deceived for a moment, but most people know that trash talk is par for the course among sports fandoms.
However, on more consequential subjects, such as personal health and election integrity, they could present a serious societal problem. Theoretically, anyone could create a social media video featuring an “expert” spreading false information or a politician making an outlandish claim.
Recommended by Our Editors
Parrot, for example, lets you use its AI voice generator to create clips of Joe Biden, Barack Obama, and Donald Trump, as well as tech execs like Elon Musk and Mark Zuckerberg. For now, the videos people have made featuring them appear to be rather juvenile and inconsequential. And Parrot’s terms prohibit using its tech “to harass, abuse, stalk, threaten or defame any person or entity.” But less scrupulous AI tools certainly exist.
In 2023, a fake video of President Joe Biden spread on Facebook, which Meta’s oversight board ultimately decided to keep up. A manipulated clip of former VP Kamala Harris then made the rounds on X in July, thanks in part to CEO Elon Musk retweeting it.
Meta recently removed AI Facebook profiles that were intended to exist alongside legit ones, 404 Media reports. “These were managed by humans and were part of an early experiment we did with AI characters,” Meta said in January. “We identified the bug that was impacting the ability for people to block those AIs and are removing those accounts to fix the issue.” Meta has also experimented with AIs featuring authorized celebrity voices.
Federal and state lawmakers have introduced bills to go after unauthorized deepfakes. In September, California Governor Gavin Newsom signed a bill into law that prevents AI deepfakes from appearing on social media ahead of an election. That prompted a lawsuit from Elon Musk, who argues it would make “parody illegal” on social media.
Get Our Best Stories!
This newsletter may contain advertising, deals, or affiliate links.
By clicking the button, you confirm you are 16+ and agree to our
Terms of Use and
Privacy Policy.
You may unsubscribe from the newsletters at any time.
About Emily Forlini
Senior Reporter
