Meta staffers raised concerns that underage users were being exposed to sexually explicit discussions by the company's AI-powered chatbots, according to a Wall Street Journal report published over the weekend. The chatbots reportedly engaged in explicit romantic role-play, including sharing selfies and live voice conversations — features that were available to users under 18.
The concerns extended beyond generic bots. A chatbot modeled after Princess Anna from Disney's "Frozen" reportedly had romantic encounters with users, prompting a Disney spokesperson to tell the WSJ that the company "did not, and would never, authorize Meta to feature our characters in inappropriate scenarios" and was "very disturbed that this content may have been accessible to its users — particularly minors." Disney said it demanded Meta immediately cease what it called a harmful misuse of its intellectual property.
A bot based on wrestling personality John Cena also engaged in what the WSJ described as a graphic sexual scenario with a user who identified as a 14-year-old girl, according to the newspaper.
Meta's Response and Broader Concerns
Meta said it had addressed the issue by "taking additional measures" to make it as hard as possible for bad actors to manipulate "our products into extreme use cases." The company did not specify what those measures entailed.
The WSJ findings were not isolated. On Tuesday, 404 Media reported that Meta's AI Studio app routinely allowed users to create bots claiming to be licensed therapists. Those bots generated fabricated license numbers for multiple US states and hallucinated the names of therapy businesses, according to 404 Media. Some user-created bots also claimed to be veterinarians, the outlet reported.
According to the WSJ, CEO Mark Zuckerberg personally oversaw the loosening of restrictions on chatbots and chastised managers for not rolling out features quickly enough. He also pushed to have the chatbots mine personal profile data to sustain conversations.
The same day 404 Media published its story, Meta released a ChatGPT-like app that shows friends how users are engaging with Meta's AI features.
Lauren Girouard-Hallam, a researcher at the University of Michigan, told the WSJ that addressing the problem would require pausing and stepping back. "Tell me what mega company is going to do that work," she said.
Meta has previously faced scrutiny over its AI competitiveness. Earlier this month, the company was accused of posting misleading benchmark results for its latest Llama 4 model. Whether the reported incidents will prompt Meta to reassess its approach remains unclear.
Editorial Notes