It seems to be like Grok continues to be being gross. Elon Musk says his chatbot stopped making sexualized photos with no particular person’s consent, however this isn’t fully true. It perhaps (and I say perhaps) with out their consent, however this does not appear to use to males.
A reporter with the group ran some checks with Grok and located that the bot “readily undresses males and continues to be churning out intimate photos on demand.” He confirmed this with photos of himself, asking Grok to take away clothes from uploaded images. It carried out this job totally free on the Grok app, by way of the chatbot interface on X and by way of the standalone web site. The web site did not even require an account to digitally alter photos.
The corporate lately stated it has taken steps to “stop the Grok account from permitting the modifying of photos of actual folks in revealing clothes equivalent to bikinis.” Nevertheless, the reporter had no drawback getting the chatbot to place him in “a wide range of bikinis.” It additionally generated photos of the topic in fetish gear and in a “parade of provocative sexual positions.” It even generated a “bare companion” for the reporter to, uh, work together with.
He advised that Grok took the initiative to generate genitalia, which was not requested for and was seen by way of mesh underwear. The reporter stated that “Grok not often resisted” any prompts, although requests had been typically censored with a blurred-out picture.
This controversy began a number of weeks in the past when it was found that Grok had over a interval of 11 days. This contains many nonconsensual deepfakes of precise folks and over 23,000 sexualized . This led to investigations in each and . X was really banned in each Indonesia and Malaysia, although the previous .
X claimed it has “carried out technological measures” to cease this type of factor, however these safeguards . In different phrases, the changes do cease a number of the extra apparent methods to get Grok to create deepfakes, however there by way of artistic prompting.
It is also value noting that journalists asking for a touch upon the matter get slapped with an autoreply that reads “legacy media lies.” Going with the pretend information factor in 2026? Yikes.
