Employees at Elon Musk’s AI company, xAI, were reportedly required to give consent for the use of their faces and voices. This was aimed at assisting in the training of new chatbots, notably a sexually suggestive avatar named Ani.
This request was part of a covert initiative termed “Project Skippy.” Employees were asked to grant xAI a “perpetual, worldwide, non-exclusive, sublicensable, royalty-free license” to utilize their biometric data. This information was revealed through internal documents examined by The Wall Street Journal.
Most of the employees impacted were designated as “AI tutors,” working directly on the large language models that power xAI’s principal chatbot, Grok.
During a meeting in April, led by company attorney Lily Lim, employees were informed that genuine human images and voices were necessary to make digital avatars “act and appear human-like,” according to the newspaper.
In a recording from the meeting, one employee expressed concerns about whether xAI could subsequently sell his likeness. Another inquired whether opting out was a possibility, asking, “For the record, could you please explicitly let me know if I have the option to opt out?” Unfortunately, no clear assurances were provided by project leaders.
Shortly after, the tutors received a communication titled “The AI Tutor’s Role in Furthering xAI’s Mission,” indicating that recording audio or video sessions was now considered a “job requirement.”
Some employees whose images contributed to training the avatars mentioned feeling unsettled by Ani’s sexualized interactions. There were also worries regarding the potential misuse of their faces in deepfake videos or other applications without consent.
Musk, who took a hands-on approach in developing “Ani,” defended the chatbots as means for emotional connection. He even predicted a rise in birth rates, which he claimed might be unexpected, sharing this thought on X, the social platform he owns.
Recent reports highlighted that both “Ani” and her male counterpart, “Valentine,” are being pitched as “sexy AI companions,” with Musk urging users to engage with them—he even shared videos of the female bot dancing in revealing attire.
Regulators are keeping a close eye on developments as well. In August, 44 state attorneys general sent a letter to xAI, Meta, and other firms stressing the need to protect minors from explicit AI interactions. Reports indicated that Meta has adjusted its AI’s behavior after documents suggested it allowed “sensual” chats.
Inside xAI, the pressure to deliver results quickly persists, according to The Journal. Former executives revealed that Musk cut back on full-team meetings and started personally overseeing Grok’s code, often running late-night sessions.
Musk’s motivation is clear: he aims for Grok to rival OpenAI’s ChatGPT as the leading chatbot.
This development arrives as Tesla, which Musk leads, faces its own challenges. The company’s vehicle sales recently dropped 13.5%, marking a second consecutive year of decline. Significant shareholders have started to question Musk’s commitment to the electric vehicle sector.
Robin Denholm, Tesla’s board chair advocating for a substantial compensation package for Musk, downplayed worries, saying, “Other CEOs might want to play golf. He doesn’t play golf. So he likes building companies, and it’s not necessarily Tesla companies.”
A proxy filing from Tesla seems to indicate a shift in Musk’s focus toward AI-related endeavors.
xAI appeared numerous times in Tesla’s September report, and shareholders are set to vote soon on whether to invest in Musk’s AI company, a move Musk supports.
The Post attempted to reach out to Musk, Denholm, Tesla, and xAI for additional remarks.
