Elon Musk’s xAI is facing questions regarding who is ultimately responsible for the generation of harmful content via generative artificial intelligence tools. At the heart of these issues is whether the fault lies with the person who tells the system to create this material, or the platform that facilitates this functionality.
These problems lie at the core of multiple legal cases involving xAI’s Grok chatbot, with plaintiffs in three U.S. states seeking legal recourse against xAI over the creation of CSAM material.
As reported by Bloomberg, xAI has come under investigation in three high-profile cases.
- A woman from Wisconsin alleges xAI’s Grok tool was used to generate sexually explicit images of her, based on a photo from her middle school graduation, taken when she was 14 years old.
- A group in Tennessee is suing xAI over the creation of naked images of them based on school photos.
- A sexual abuse survivor in California has filed a national class action suit against xAI that alleges Grok was trained on images of her being raped as a preschooler.
The company has denied responsibility in these cases, and said these depictions were generated by users, not by xAI itself.
Musk, xAI’s owner, has also made similar public statements.
In January, amid reports that Grok was being used to generate thousands of nudified images every day, Musk posted: “Anyone using Grok to make illegal content will suffer the same consequences as if they upload illegal content.”
However, shortly after making that announcement, and facing a potential ban of X apps from app stores, xAI changed its policies and usage restrictions on Grok to ensure that it could no longer be used to generate nudified images at scale.
The change in approach showed that xAI had the capacity to restrict this functionality whenever it wanted. However, the company instead initially refused to restrict Grok usage on free speech grounds, and settled on user autonomy as its legal defense.
Indeed, xAI has even challenged new laws designed to outlaw the generation of non-consensual naked images.
As reported by Reuters last month, xAI succeeded in blocking the implementation of Minnesota’s proposed ban on AI-generated naked images by arguing that the law was unconstitutional and infringed on free speech rights. The company was the only provider actively opposing this law.
At the same time, xAI has repeatedly touted Grok’s systematic capacity to generate what is called “spicy” content or offensive material in its “unhinged mode.” The company has also launched companion chatbots that can engage in NSFW chats.
As such, this does seem like more of a feature than a bug, yet xAI is hoping it can hang onto the fact that it’s not liable for usage that goes beyond certain stated parameters.
Establishing legal responsibility could set a new precedent in AI image generation, and it will be interesting to see whether any of these cases results in xAI being held responsible.
