Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Elon Musk’s xAI has restricted the image-generation function on its Grok AI model in response to an outcry over the proliferation of fake sexualised images on the X social media platform.
The billionaire entrepreneur came under scrutiny last week after thousands of people began using Grok to generate sexualised deepfakes of women and, in some cases, children — sharing them on X as well as the separate Grok app, both of which are run by Musk’s xAI start-up.
After facing pressure from governments around the world, including bans and threats of legal action, the company late on Wednesday announced that it had “implemented technological measures” to limit Grok from generating certain sexualised images.
It marked a rare retreat by the world’s richest man, who has pushed for his AI products to have fewer content “guardrails” than competitors such as OpenAI and Google in line with his “free speech” ideals, but also in a bid to drive downloads of his Grok app, according to insiders.
xAI said in a post on X that Grok would no longer allow “the editing of images of real people in revealing clothing such as bikinis”.
Generating images of “real people in bikinis, underwear and similar attire via the Grok account and in Grok in X” will also be blocked in jurisdictions where it is illegal, it added.
The climbdown came after the UK government announced it would accelerate the enforcement of new powers making it a criminal offence to create non-consensual intimate images, including with AI.
Earlier on Wednesday, UK Prime Minister Sir Keir Starmer said that Musk’s social media platform had indicated to government officials that it was acting to comply with the country’s online safety laws by blocking the generation of non-consensual sexual imagery.
“We have made clear to X that these images are illegal, reprehensible and need to be dealt with,” Starmer said. “I have been informed that X is acting to ensure full compliance with UK law. If so, that is welcome, but we’re not going to back down — they must act.”
Musk had last week pushed back against threats by the UK government to ban the platform over the issue, saying they wanted “any excuse for censorship” and arguing that X was being singled out where other rivals offered similar capabilities.
But in a post on X on Wednesday, Musk said: “When asked to generate images, it 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 that he was “not aware of any naked underage images generated by Grok. Literally zero.”
The issue has prompted threats of fines and bans in the EU, UK and France, and the announcement of an investigation by Britain’s Ofcom regulator. Grok has also been banned in Indonesia and Malaysia.
The California attorney-general on Wednesday said the US state had opened an investigation into Grok and xAI “over undressed, sexual AI images of women and children”.
A spokesperson for the European Commission earlier on Wednesday said the EU “took note of the additional measures X is taking to ban Grok from generating sexualised images of women and children” and added that “should these changes not be effective, the commission will not hesitate to use the full enforcement toolbox of the DSA”, referring to its Digital Services Act, which polices content online.
Last Friday, xAI said it was limiting the use of its Grok image generator to paid subscribers only. On Wednesday, X wrote that this decision “adds an extra layer of protection by helping to ensure that individuals who attempt to abuse the Grok account to violate the law or our policies can be held accountable”.
It added: “We take action to remove high-priority violative content, including Child Sexual Abuse Material (CSAM) and non-consensual nudity, taking appropriate action against accounts that violate our X Rules.”
As of late on Tuesday, Grok appeared to be ignoring requests from a number of X users across the globe to generate sexualised deepfakes, however some were still being generated, according to data provided by the Institute for Strategic Dialogue think-tank.
The number of requests for non-consensual deepfakes exploded earlier in January, particularly after Musk shared a fake image of himself in a bikini.
It also came as xAI has experienced turnover in its safety teams. In the first week of December, xAI’s head of product safety Vincent Stark left the company, along with one of its top AI safety researchers Norman Mu and Alex Chen, who oversaw the fine-tuning of Grok’s personality and behaviour.
Additional reporting by Barbara Moens in Brussels
Credit: Source link
