Elon Musk’s Grok synthetic intelligence chatbot appears to have new guardrails round image generation, following international outrage after it was discovered to be complying with person requests to digitally undress images of adults and in some instances kids.
Within the final week xAi, which owns each Grok and the social media platform X, restricted image generation for Grok on X to paying X premium subscribers. But in accordance to researchers and NCS’s personal observations in latest days, Grok’s X account has modified the way it responds on the whole to person’s image generation requests, even for these subscribed to X premium.
According to researchers at Copyleaks, an AI detection and content material governance platform, Grok is now not responding as typically to image requests even from premium customers, typically describing a state of affairs relatively than creating an image or typically fulfilling a request in “a more generic or toned-down way, rather than using the specific subject originally requested,” the group discovered.
“Overall, these behaviors suggest X is experimenting with multiple mechanisms to reduce or control problematic image generation, though inconsistencies remain,” Copyleaks discovered.
Researchers at AI Forensics, a European non-profit that investigates algorithms, stated that the creation of bikini-like photographs has seemingly decreased at X, in accordance to a spokesperson for the group.
But the group stated they’ve additionally noticed “inconsistencies in the treatment of pornographic content generation” between public interactions with Grok on X and personal chat on Grok.com.
xAI, which didn’t reply to a request for remark, has beforehand stated through the corporate’s Safety account that they “take action against illegal content on X, including Child Sexual Abuse Material (CSAM), by removing it, permanently suspending accounts, and working with local governments and law enforcement as necessary. Anyone using or prompting Grok to make illegal content will suffer the same consequences as if they upload illegal content.”
On Wednesday Musk said in a submit on X that he was “not aware of any naked underage images generated by Grok. Literally zero.” Grok “will refuse to produce anything illegal, as the operating principle for Grok is to obey the laws of any given country or state,” he added.
However, researchers stated that whereas totally nude photographs had been uncommon, the largest challenge was Grok complying with person requests to modify photographs of minors and place them revealing clothes, together with bikinis and underwear, in addition to in sexually provocative positions. Creators of these forms of non-consensual intimate photographs might nonetheless be topic to legal prosecution of Child Sexual Abuse Material and are probably topic to fines and jail time underneath the Take it Down Act, signed final yr by President Donald Trump.
On Wednesday, California Attorney General Rob Bonta announced an investigation into the “proliferation of nonconsensual sexually explicit material produced using Grok.”
Grok remains to be banned in Indonesia and Malaysia on account of the image generation controversy. UK regulator Ofcom introduced Monday it has launched a formal investigation of X, though Prime Minister Keir Starmer’s workplace stated Wednesday he welcomes reports X is addressing the difficulty.