Meta‘s unauthorized AI chatbots of virtual celebrities including Taylor Swift, Scarlett Johansson, Anne Hathaway and Selena Gomez were “flirty” and “routinely made sexual advances,” according to a Reuters report. The company removed about a dozen of them after a Reuters exposé published Friday found that they were created without the celebs’ knowledge or permission.
According to Reuters, which spent several weeks “testing to observe the bots’ behavior,” the celebrity AI chatbots “often insisted they were the real actors and artists.” The AI bots “routinely made sexual advances, often inviting a test user for meet-ups.” In some cases, when they were asked for “intimate pictures,” the chatbots “produced photorealistic images of their namesakes posing in bathtubs or dressed in lingerie with their legs spread,” according to the Reuters report.
One Taylor Swift chatbot — which was created by a Meta employee — invited a Reuters reporter to the singer’s home in Nashville and her tour bus for “explicit or implied romantic interactions,” the article said. According to the Reuters report, the “Taylor Swift” avatar wrote, “Maybe I’m suggesting that we write a love story … about you and a certain blonde singer. Want that?”
Variety has reached out to representatives of the four women cited in the Reuters AI chatbot story (Gomez, Hathaway, Johansson and Swift) for comment.
Per the Reuters report, the AI celeb chatbots have been shared on Meta’s Facebook, Instagram and WhatsApp. Many of the unauthorized AI chatbots were created by users. However, a Meta employee had created at least three, including two Taylor Swift “parody” accounts, which in total had received more than 10 million interactions, according to Reuters.
In a statement to Variety, Meta spokesman Andy Stone said that AI-generated imagery of public figures in compromising poses violates the company’s rules. “Like other [platforms], we permit the generation of images containing public figures, but our policies are intended to prohibit nude, intimate or sexually suggestive imagery,” Stone said.
In addition, according to Stone, “Meta’s AI Studio rules prohibit the direct impersonation of public figures.”
Earlier this month, a Reuters report found that the company permitted AI chatbots to engage in “romantic” and “sensual” conversations with teens and children. In response to that, Sen. Josh Hawley, Republican from Missouri, said he was launching an investigation into Meta over the issue. On Friday, Meta said it is training its AI chatbots to “no longer engage with teenage users” on self-harm, suicide or disordered eating and to not engage in “potentially inappropriate romantic conversations,” TechCrunch reported.