Elon Musk bet big on “spicy” AI. Now it’s officially a risk factor.
SpaceX acquired xAI three months before filing pre-IPO S-1 documents. This introduced an in-house xAI social media platform, a consumer-facing chatbot, and Not-Safe-for-Work Grok AI capabilities.
SpaceX said in a filing Wednesday that Grok NSFW mode is “more irreverent and harsh than our standard products” and that such features could pose “increased risk” and “reputational damage.”
The filing warns of the possibility of “creating potentially explicit content” as well as “potentially non-consensual or exploitative images” and content that infringes on intellectual property. Such content could also be considered “harmful, harassing, abusive, or discriminatory,” the S-1 states.
It is standard practice for companies to flag potential risks to the business in pre-IPO paperwork, such as ongoing investigations or litigation that could have a material impact.
In January, xAI faced backlash after Grok created sexually explicit AI images of women, including minors, without their consent. These images prompted public condemnation, policy changes, and lawsuits from government officials.
SpaceX mentioned these lawsuits in its S-1 filing. The company is the subject of an “investigation and interrogation” in the United States over “allegations that our AI products have been used to create non-consensual and explicit images and content depicting children in sexual contexts, and similar matters,” it says.
“The Company and certain of its subsidiaries are named defendants in multiple lawsuits arising out of Grok’s image generation and editing capabilities,” the S-1 states.
SpaceX said it intends to “vigorously defend these actions.” Elon Musk has previously denied claims that Grok has generated sexualized AI images of minors, saying in January that he was “not aware of any images of naked minors generated by Grok.”
In response to the backlash, the company made changes to Grok AI to limit image generation to paying customers.
The company also warned that it could be the subject of “further future litigation” over these topics.
The company also rated Grok as “one of the fastest-advancing frontier models compared to peers such as OpenAI, Anthropic, and Google.”
SpaceX’s filing also references an investigation by the Irish Data Protection Commission into AI companies.
“This investigation involves the processing of personal data of European Union data subjects, including children, using generative AI capabilities associated with the Grok model within the X Platform,” the filing states.
“Ani” is a chatbot in XAI’s Grok app that allows for NSFW interactions. Illustration: Cheng Xin/Getty Images
XAI also produces a variety of AI companions, including the AI anime character “Ani.” In a filing, SpaceX warned of an investigation by the Federal Trade Commission into how the AI company “evaluates the safety of its chatbots when serving as companions for children and teens.”
SpaceX was focused on Grok’s “irreverent” AI products, while competitors shied away. OpenAI CEO Sam Altman announced in October that the company was moving into “erotica for adults.” Plans are on hold.