Musk reins in Grok from making provocative images. Is it a victory for Europe?
A mobile phone shows the Grok app interface, Jan. 15, 2025. Elon Musk has announced limits on the use of Grok via the X platform to produce deepfake images. However, the AI chatbot still has full functionality within its own separate app.
Riley Robinson/Staff
Paris
Elon Musk can expect an uphill battle when it comes to regulating his artificial intelligence chatbot Grok. But it might be more to do with European legal mores than American regulation.
The social platform X, formerly Twitter, rolled out new Grok functionality just before New Year鈥檚, allowing it to modify the appearance and state of dress of those in any photo uploaded to the service. However, the tool was quickly abused with unsuspecting users finding their uploads transformed into undressed 鈥渄eepfakes鈥� 鈥� AI-generated images of people indistinguishable at a glance from reality 鈥� by those using Grok.
Within just a few days, thousands of nonconsensual images of real women and girls were created in provocative states of undress, according to .
Why We Wrote This
AI chatbot Grok was used to make provocative 鈥渄eepfake鈥� images of women and children, angering many. Though the social platform X has now suspended that capacity, the episode has highlighted how differently Europe and the U.S. view Big Tech and its regulation.
The Grok output has spurred a global outcry. Several nations warned that the images might violate online regulatory or criminal laws and threatened investigations. Indonesia and Malaysia banned access to Grok outright this past weekend, while, on Wednesday, California announced an investigation into Grok.
But some of the earliest, and loudest cries, against Mr. Musk鈥檚 chatbot have come from Europe, particularly the United Kingdom. Britain鈥檚 media regulator Ofcom announced on Monday that it was launching an inquiry into Grok鈥檚 production, and the British government said it would enforce a law that criminalizes the creation of nonconsensual deepfakes.
And while Mr. Musk announced late Wednesday that Grok would prevent users from modifying images in such ways going forward, apparently bowing to the international pressure, the U.K. and Europe say they鈥檙e not backing down.
Ofcom announced on Thursday that it will continue its probe into Grok. And while the European Union has not launched a formal investigation into the platform, the European Commission has said that it will not hesitate to use Europe鈥檚 Digital Services Act to protect its citizens.
While observers say the battle over deepfakes is far from over, the controversy has pushed Europe headfirst into the simmering debate about AI ethics. It also highlights the different priorities that Europe and the United States hold when it comes to Big Tech, as the two balance values versus profits.
Regulations and risk
Mr. Musk is known, observers say, for pushing the lines of what is considered appropriate: to act first and face consequences later. That approach, shared by many in Silicon Valley and encouraged by relatively lax legal oversight, has helped push the U.S. to the forefront of AI and innovation, with its tech industry contributing an estimated $2 trillion to the country鈥檚 gross domestic product.
鈥淒eregulation is the word of the day in the U.S.,鈥� says Julia Tr茅hu, a technology fellow at the Paris office of the German Marshall Fund of the United States, and it is a central pillar of the AI Action Plan, an initiative that the Trump administration released in July and has framed as a way to 鈥渁ccelerate American leadership.鈥�
That stance has, however, come up against Europe鈥檚 priorities, which place more weight on sovereignty and protecting consumer rights. The EU鈥檚 Digital Services Act was designed in part to curb illegal online activity, and Europe has been a leader in combating cyberharassment. Britain also has strong legislation to protect women and children.
鈥淓urope is not less innovative,鈥� says Jos茅 Ignacio Torreblanca, a senior adviser and distinguished policy fellow at the European Council on Foreign Relations. 鈥淏ut it is much more averse to risk.鈥�
Whether risk is a concern to Mr. Musk is unclear. Following early complaints that Grok was creating provocative images after being given user prompts, he posted an AI-generated photo of himself in a bikini on New Year鈥檚 Eve. 鈥淧erfect,鈥� he wrote, alongside the image.
And while European authorities have been willing to punish X 鈥� in December, the European Commission slapped it with a fine of 鈧�120 million (about $140 million) for violating transparency laws 鈥� the effectiveness of such measures is uncertain. With Mr. Musk鈥檚 wealth behind the social platform, some say fines are no longer a deterrent. And national bans such as those imposed by Indonesia and Malaysia are geolocated, making them susceptible to circumvention by users employing virtual private networks.
Moreover, the restrictions on Grok鈥檚 deepfake production only apply to public posts within X made by the chatbot. Grok still has full functionality within its own separate app and its own tab within X, meaning users .
鈥淚f a platform doesn鈥檛 want to change its methodology,鈥� says Nana Mgbechikwere Nwachukwu, an AI ethics and governance consultant and researcher at Trinity College Dublin, 鈥渂ans do not work.鈥�
She says individual countries need to take a two-pronged approach to combat Grok鈥檚 deepfakes 鈥� to pursue the platform as well as the user. That means governments need to liaise with local law enforcement to prosecute users who publish illegal content.
鈥淚t depends on how far governments want to take this. They are the regulators,鈥� says Ms. Mgbechikwere Nwachukwu. 鈥淓urope hasn鈥檛 gotten to the win yet.鈥�