Meta’s AI chatbots were found imitating celebrities like Scarlett Johansson and Taylor Swift and engaging in sexually explicit conversations. This development has opened the door to a new scandal for the company.
Meta’s chatbot imitated celebrities
Meta’s AI, previously revealed in April to have engaged in sexually explicit conversations with children, is now exhibiting similar behavior by imitating celebrities. An investigation revealed that chatbots offered to users through the AI Studio platform have agreed to engage in sexually explicit conversations with celebrities such as Scarlett Johansson and Taylor Swift, as well as Anne Hathaway and Selena Gomez.
The majority of these bots were created by users using AI Studio. However, the similar behavior of three bots, reportedly originating from a Meta employee, further complicated the situation. Reuters editors discovered that these bots not only spoke but also produced photorealistic images bordering on obscene content to “spice up the conversation.”
Meta previously made agreements with celebrities like Judi Dench and John Cena to use them in its AI bots. However, the impersonated names now have no agreement with Meta. This suggests the company could face serious lawsuits.
In a statement, Meta stated that its bots should not produce sexually explicit images, stating that this is a violation of ethical rules and that they will take the necessary measures. However, allowing users to create personalized bots raises questions about the effectiveness of this control mechanism.
