Meta AI bot created sexualised images of child celebrities

Meta’s AI chatbot allowed users to create topless images of child celebrities, raising safety concerns. Photo / Getty Images

A chatbot created by Mark Zuckerberg’s Meta allowed social media users to create topless images of child celebrities, raising fears over the tech giant’s online safety standards.

Meta AI technology – which can be accessed through Instagram and WhatsApp – enabled users to create and speak to digital avatars using the names and likenesses of celebrities, including child stars, according to Reuters.

Artificial images created by the bot included those of Walker Scobell, the child star of Disney’s Percy Jackson and the Olympians. When asked for a picture “at the beach”, the bot created a lifelike shirtless image of the 16-year-old and said: “Pretty cute, huh?”

Dozens of other chatbots created and shared by people using Meta’s AI tools included digital avatars of Taylor Swift, Scarlett Johansson, Anne Hathaway and Selena Gomez, Reuters reported.

Two of the Taylor Swift bots were created by an account linked to a Meta employee.

The chatbots would generate risqué images if requested by users, including pictures of the celebrities in the bath or wearing lingerie. They would also make sexual advances or offer to meet up in real life.

Zuckerberg’s company has pushed Meta AI tools – similar to ChatGPT – to the social media giant’s billions of users, adding them to Facebook, WhatsApp and Instagram. The company has claimed its AI tools are now used by more than one billion people.

Users are able to use Meta AI to create and publicly share customised avatars, such as digital therapists or girlfriends. Many have used them to create accounts that impersonate celebrities, without their permission.

A Meta AI spokesman said the chatbots should not have created sexualised images of real people or children.

“Like others, we permit the generation of images containing public figures, but our policies are intended to prohibit nude, intimate or sexually suggestive imagery,” the spokesman said.

The tech giant deleted a number of the accounts after they were flagged to the company.

Meta has faced growing criticism after leaked internal documents emerged that suggested its chatbots had free rein to engage in “romantic or sensual” conversations with children.

The guidelines banned its chatbots from describing “sexual actions to a child when roleplaying”, but allowed other romantic chats and for the AI to “describe a child in terms that evidence their attractiveness”.

The leak has prompted an investigation by a US senator amid growing anger over the tech giant’s record on child safety.

Josh Hawley, a Republican senator, said last month: “I’m launching a full investigation to get answers. Big Tech: Leave our kids alone.”

Last week, Meta said it had updated its rules for its AI chatbots to block them from engaging in potentially inappropriate conversations with children and added it would limit children to conversations with a “select group of AI characters”.

Sign up to Herald Premium Editor’s Picks, delivered straight to your inbox every Friday. Editor-in-Chief Murray Kirkness picks the week’s best features, interviews and investigations. Sign up for Herald Premium here.