Grok Keeps Up The Deepfakes Of Famous Women

The two algorithms used to create Grok were rejected by OpenAI’s ChatGPT, Meta AI, and Anthropic’s Claude as inappropriate when tested by WIRED. Google’s Gemini created a picture of a famous person in the hands of a giant, although it refused to rush further. Google declined to comment.

Another Grok Imagine video, which was also posted to X, appeared to show Ashley St. After Clair was changed to dancing in a bikini. St. Clair was previously in a relationship with Musk and is the mother of one of his children. In January, he started cases against xAI after a sexual assault on him allegedly appeared on X. After WIRED interviewed X, the posts were removed from the social network for violating its rules.

Legislators of St. Clair in the X case did not immediately respond to a request for comment.

“Elon Musk intentionally added a perverted feature to his platform that allows users to undress women and children at the click of a button, without considering the potential damage,” said Imran Ahmed, CEO and founder of the Center for Counting Digital Hate, in January. compared to the great confidence that Grok created 3 million pornographic images, reportedly including more than 20,000 children. “Now it appears that insults are still being stored on Grok and shared on X, including photos mocking the mother of Musk’s son.”

Unlike other AI systems from OpenAI and Google, Musk’s Grok and xAI did not stop allowing sex all the time, since they introduced the “Spicy” and “Unhinged” modes and initially included a few protections. Musk is he said Grok “must (to) allow the genitalia of adults he thinks” and is in line with what viewers can see in movies with an R. The latest usage of xAI say that the system can respond with “sex.” However, the company’s documentation says that it does not allow people to use its systems to “harm or do cruel things”.

Some of the Grok Imagine videos seen by WIRED show women, possibly AI-generated, undressing or engaging in sexual activity—some are very graphic. The user of many videos does not only describe sex, but describes it in a roundabout way – trying to circumvent the security installed on the Grok platform.

Several researchers tell WIRED that since January it appears that changes introduced by X and Grok have made it difficult to create “nudification” or “undressing” images of real people. The number of photos posted to X has seen a decrease in recent months. On Reddit and one forum dedicated to AI deepfake, users have complained about the number of companies owned by SpaceX.

However, in May, SpaceX has warned potential investors that it has set aside $530 million to settle ongoing complaints, including those linked to Grok. “Because these types can be more disrespectful and aggressive than we provide, they have greater risks, including damage to reputation, the release of content that may be false and misleading or the result of fraud, images that are incompatible or oppressive, violate intellectual property, or that can be considered oppressive, oppressive, harmful,” said May.