Elon Musk bet big on “spicy” AI. Now, it is formally a threat issue.
SpaceX acquired xAI three months earlier than it filed its pre-IPO S-1 paperwork. That introduced in-house xAI’s social media platform, shopper chatbot — and not-safe-for-work Grok AI options.
Within the Wednesday submitting, SpaceX mentioned that such options might pose “heightened dangers” and “reputational hurt” as a result of the Grok NSFW modes are “extra irreverent and harsher than our customary choices.”
The submitting warned in regards to the potential “technology of doubtless express content material,” in addition to “potential nonconsensual or exploitative imagery” and content material that infringes on mental property. Such content material may be considered as “dangerous, harassing, abusive, or discriminatory,” the S-1 learn.
It is customary follow for corporations to flag potential dangers to its enterprise in its pre-IPO paperwork, together with any ongoing investigations or lawsuits that might have a fabric influence.
In January, xAI confronted backlash after Grok created non-consensual sexualized AI images of women, including minors. The pictures sparked public condemnations from authorities officers, coverage modifications, and lawsuits.
SpaceX referenced these lawsuits within the S-1 submitting. The corporate is topic to “investigations and inquiries” within the US regarding “allegations that our AI merchandise had been used to create nonconsensual express photos or content material representing kids in sexualized contexts, and related issues,” it learn.
“The Firm and sure subsidiaries have been named as defendants in a number of lawsuits arising from Grok’s image-generation and modifying options,” the S-1 reads.
SpaceX mentioned it intends to “defend itself vigorously in these actions.” Elon Musk has beforehand pushed again on claims that Grok generated sexualized AI photos of minors, stating in January that he was “not conscious of any bare underage photos generated by Grok.”
The corporate made modifications to its Grok AI following the backlash, limiting image generation to paying prospects.
The corporate additionally warned that it could be topic to “extra litigation sooner or later” surrounding such matters.
The corporate additionally framed Grok as “among the many fastest-advancing frontier fashions relative to friends, together with OpenAI, Anthropic, and Google.”
The SpaceX submitting additionally referenced the Irish Data Protection Commission’s inquiry into the AI firm.
“This inquiry includes the processing of private information of European Union information topics, together with kids, utilizing generative AI performance related to the Grok mannequin throughout the X platform,” the submitting learn.
illustration by Cheng Xin/Getty Photos
XAI additionally produces quite a lot of AI companions, like its AI anime character, Ani. Within the submitting, SpaceX warned about an inquiry from the Federal Commerce Fee about how AI corporations “have evaluated the security of their chatbots when appearing as companions to kids and youths.”
Whereas SpaceX had doubled down on Grok’s “irreverent” AI choices, its opponents have shied away. OpenAI CEO Sam Altman introduced a push into “erotica for adults” in October; the plans are on hold.
