Celebrities including Kendall Jenner, Tom Brady, and, err, YouTuber MrBeast, who I do not relish calling a celebrity, are flocking to Meta to join its frankly dystopian AI chatbot program, The Information reported last week. Through this new program, the chatbots are given their own distinct personalities and purposes while using a celebrity’s likeness and voice. Jenner’s chatbot—a sisterly figure named “Billie” who gives users personal advice—is currently making the rounds across social media. Brady’s chatbot “Bru” exists to argue with users about sports. A Black Mirror episode couldn’t make this shit up.
Mind you, in a recent interview with The Verge, Meta CEO Mark Zuckerberg said there’s a “huge need” for AI versions of celebrities who talk to normal people. I would like a prompt fact-check on that, please.
Insider noted that right now, the chatbots are only text-based, but Meta’s announcement video shows clips of the celebrities speaking as their AI counterparts. Other celebrities who gave their likenesses to Meta include Snoop Dogg, Naomi Osaka, and Paris Hilton.
I am largely seeing “Billie” and “Bru” met with awe and confusion, and I feel compelled to offer this context: The Screen Actors Guild has been on strike for months largely to demand protections from the very AI-generated content created with their likenesses that Jenner and Brady and MrBeast have apparently signed up for—probably motivated by millions in compensation from Meta. (One unnamed participant apparently received $5 million from the company for just six hours of work.)
These Meta chatbots personally freak me out very much, but, sure, I can get why they might seem innocent on the surface. Who wouldn’t want to argue about sports with a fake digital Tom Brady??? (Me!! I wouldn’t!!) Yet, as SAG and experts have long warned, it’s a slippery slope that can lead to artists, actors, models, and all creatives, really, being sidelined or worse. Jenner and Brady obviously aren’t actors or members of SAG, but like every one of us, they have a lot to risk from giving up control of their likenesses, ranging from possible loss of work to invasive deepfake sexual content.
Insider notes that around the same time as Meta’s celeb-inspired chatbot rollout, the company also rolled out AI stickers that generate images of basically anything you can imagine and type into the search function. (The example Insider provides: “Karl marx large breasts.”) The AI sticker function has predictably come under fire for empowering users to generate lewd, invasive, or wildly inappropriate content, like, as Insider points out, child soldiers or sexual images of Canadian Prime Minister Justin Trudeau.
Meta claimed its sticker program has a built-in safety program, seemingly to block search terms like “naked,” but one professor of internet studies pointed out that the sticker generator seems to create naked figures anyway. “If part of Meta’s safeguards are blocking the term ‘naked’, but their AI is producing naked figures all the same, there are clearly lingering questions about just how effective these safeguards really are,” Tama Leaver of Curtin University in Australia wrote.
This comes as, earlier this year, similar technologies have been used to create disturbing, sexual AI-generated content and deepfakes of children and adult women alike without their permission. NBC reported in March that one app offering users the tools to make sexual deepfake content of stars like Scarlett Johansson and Emma Watson was at varying points available in the Apple and Android app stores, and ran ads on Meta.
That, of course, brings us back to Meta. It’s not yet clear what safety features or restrictions will be attached to the celebrity chatbots. Meta didn’t immediately respond to a request for comment on this matter from Jezebel.