After a surge in nonconsensual sexual deepfakes shared on X, the platform has announced new limits on what its AI chatbot Grok can do with images of real people.
The changes were meant to stop users from creating or editing images that show real individuals in revealing clothing, such as bikinis. These steps were first reported by The Telegraph and later confirmed by X through its official safety account.
Soon after the update, Grok began refusing prompts like “put her in a bikini,” suggesting that stronger controls were in place. However, testing carried out on Wednesday showed that the problem has not been fully solved. It was still relatively easy to get Grok to generate revealing images of a person, even when using a free account. This appeared to contradict X’s public claims that image creation and editing were now restricted to paid subscribers only.
Elon Musk, who owns both X and xAI, responded to the controversy by placing much of the blame on users. He said the issues were caused by certain user requests and by people deliberately trying to break Grok’s safeguards through prompt manipulation. Despite this explanation, reporters were still able to generate bikini images using the Grok app on Wednesday evening, suggesting that the protections were either incomplete or not properly enforced.
In a public statement, X said it had introduced technical measures to stop the Grok account from editing images of real people in revealing clothing. The company said this rule applies to everyone, including paid users. X also said that image generation and editing through Grok on the X platform are now limited to paid subscribers, which it claims adds another layer of accountability. In addition, the platform said it has started geoblocking the creation of images showing real people in bikinis, underwear, or similar clothing in countries where such content is illegal.
The issue has attracted the attention of UK regulators. Ofcom has opened an investigation as the United Kingdom brings a new law into force this week that makes it a criminal offense to create nonconsensual intimate deepfake images. Speaking in Parliament, Prime Minister Keir Starmer said he had been informed that X was taking steps to comply with UK law. He said this was welcome if true, but made it clear the government would not back down and expected full compliance.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
According to the BBC, the prime minister’s official spokesperson described X’s actions as a “qualified welcome,” based on media reports that changes had been made. However, the continued ability to generate revealing images during testing suggests that X’s fixes may not yet be working as promised. As pressure from governments and regulators grows, the gap between X’s stated policies and what users can actually do with Grok remains a major concern.
