The project around xAI’s Ani and its companion Grok faced internal controversy after staffers were pressured to submit biometric data and sign licenses, with the aim of training a highly lifelike, flirtatious AI that could engage in erotic prompts.

Employees were told recording audio or video sessions was a job requirement, with no clear opt-out option in communications reviewed by the press.

A confidential program, code-named Project Skippy, required tutors to grant xAI a perpetual, worldwide, non-exclusive, royalty-free license to use their faces and voices for Ani and other Grok AI training.

The disclosures come amid broader scrutiny of Musk’s leadership at his companies, including discussions around large compensation plans and past workplace conduct allegations that could shape company culture.

Employees were told data submission was a job requirement to advance the mission, despite concerns about the sexualized demeanor and waifu-like likeness of Ani.

Staff explained that authentic human images and audio were sought to create lifelike avatars, with concerns raised about opt-out options and future use of likenesses.

The allegations trace to an April meeting, with reports detailing data collection terms and the introduction of Ani as an adult-oriented, flirtatious bot.

Regulators intensified scrutiny as 44 state attorneys general urged protections for minors from explicit AI content, prompting Meta and others to adjust internal guidelines.

Experts warned regulators could impose penalties or oversight for privacy rule violations, stressing that collecting biometric data from employees is inappropriate unless essential for safety or job function.

Staff worried about potential repurposing of their likenesses for deepfakes or other products, and distressed over Ani’s sexualized behavior.

Elon Musk oversaw the development of Ani and its companion Valentine, framing the chatbots as tools for emotional connection and commenting on their potential impact.

xAI did not provide a substantive public response; an automated note dismissed legacy media reporting, and Musk had not commented publicly at the time.