A review of the EU’s Artificial Intelligence safety regulations saw no evidence produced showing that X’s Grok chatbot produced material depicting child sex abuse.
The European Parliament and Council agreed on May 7 to ban so-called “nudifier” apps as part of their reforms of the bloc’s Artificial Intelligence Act.
Speaking on the deal, parliament co-rapporteur Arba Kokalari MEP claimed that the changes were in response to a controversy surrounding explicit visual content produced by the Grok chatbot last month.
Michael McNamara MEP — the other co-rapporteur for the review — however, appeared to reject the suggestion that EU lawmakers had found fault with Grok, telling Gript that neither the Council nor Parliament raised any evidence that Grok had produced child sex abuse material during the review.
This is despite repeated claims from lawmakers that the platform had produced such illegal content.
“To my knowledge, nobody alleged that Grok generated child sex abuse material in the negotiations,” he told Gript.
He added that he did not know whether Grok provides such explicit material, adding that the legislation was not specifically targeted at Musk’s flagship AI.
McNamara also clarified that the ban will be limited to realistic images of non-consenting individuals that depict “genitals, pubic area, anus, exposed buttocks or exposed female breasts, nipples or areolae”, or “sexually explicit activity”.
Many of the Grok-generated images that resulted in the January controversy, however, do not appear to fit into this framework, with much of the controversy instead centring around the AI putting politicians and public figures in risque clothing, such as bikinis.
Asked whether this would mean that many of the controversial images generated by Grok that depicted individuals in bikinis without their consent, McNamara conceded that this would indeed be the case.
“I mean, as long as their intimate parts are not exposed, that is the case, yes,” he said, clarifying that many of the finer details would be “for the courts to determine”.
The revelation follows claims by multiple bodies in Brussels that Grok had been used to generate child sex abuse material which, as defined under Irish law, requires an explicit depiction of a child viewing or engaging in sexual acts, or of their genitals.
Further restrictions on the non-consensual sharing of images depicting a person in their underwear have also been introduced as part of recent Irish government reforms known as “Coco’s Law”, though these would also not necessarily cover Grok’s “bikinification” of individuals.
This has not stopped European lawmakers from claiming X had helped produce illegal content.
Amongst those making claims that child abuse content was produced by X include McNamara’s Renew Europe group colleague, Billy Kelleher, with the Fianna Fáil MEP telling the European Parliament in January that Grok had indeed help generate such material.
“[W]hen there was high fives at the headquarters of X and Grok over the last number of weeks celebrating the fact that they had so many hits because of the AI-generated deepfakes of sexually explicit material, it sends a very chilling message for me as an elected public representative in this Parliament and not a faceless bureaucrat or not a faceless tycoon who sits behind the paywall of X promoting such visual, sexualised content,” he said in Strasbourg.
“I find it deeply offensive that we’re in a position here where we are now saying that protection of free speech requires us to allow Elon Musk and others with large social media platforms to promote soiled child sexual abuse imagery. It simply doesn’t make sense from any perspective to come into this Chamber and say that Elon Musk is a defender of free speech.”
A European Commission spokesman also repeated such a claim, telling journalists that it was a “fact” Grok was producing “explicit sexual content” of “childlike” individuals.
“This is illegal,” he said.
Elon Musk, as well as his Grok and X products, have remained under continuous scrutiny from EU lawmakers as a result of the tech billionaire’s outspoken opposition to European speech controls.
X was the first platform to be fined under the Digital Services Act late last year, with the Commission ordering the platform to hand over €120 million over claims it had not properly given European academics enough access to the platform’s sensitive data.
A further European Commission review of X’s operations — including the activities of its Grok AI — remains underway.