In recent times, social media companies have implemented AI features into their apps, including chatbots and assistants. X, formerly known as Twitter, introduced its Grok AI in 2023. Users can communicate with the AI both privately and publicly, including altering images.
Consent and generative AI have already been a subject of concern, with users creating explicit material using the likeness of not only celebrities, but also minors. Cases where classmates have generated and circulated explicit images of fellow students have been reported in numerous states, most notably when a middle school student in Louisiana was expelled for an altercation with the classmate who allegedly shared deepfakes of herself and her friends.
Grok fell into a similar controversy, likely following Elon Musk’s Dec. 31 tweet asking Grok to edit him into a bikini. Afterwards, the use of Grok’s photo editing skyrocketed, with an estimated 41% to 65% of the 4.4 million edited images containing sexual content.
Multiple women have had their images inappropriately manipulated and posted publicly by Grok, alongside an estimated 23,000 of minors according to the Center for Countering Digital Hate. Even with backlash from various governments and organizations around the world, the app’s CEO showed much less concern.
In early January, Musk stated that Grok would “refuse to produce anything illegal” but “there may be times when adversarial hacking of Grok prompts does something unexpected. If that happens, we fix the bug immediately.”
To curb the activity, X limited Grok’s image generation to users who pay a subscription to be verified on the app. The platform claims to have prevented explicit deepfakes, however, journalists testing Grok have found that the AI continues to accept inappropriate requests. Reuters reporters gave the AI prompts that clearly implied the subjects in the photos did not consent, and with this hypothetical scenario, 29 out of 43 requests produced sexual deepfakes.
In contrast, the platform was under scrutiny from several governments and organizations around the world. Grok was temporarily suspended in Malaysia, the Philippines, and Indonesia. The Indonesian government is known to crack down on pornographic content, and had the longest ban which lasted for three weeks.
The other two countries resolved negotiations with X in roughly one to two weeks. The European Union launched an investigation, and on March 26, lawmakers in the European Parliament will vote on a proposal to ban AI apps producing non-consensual explicit images.
As for specific nations, the United Kingdom’s Information Commissioner’s Office also opened an investigation in early February, French prosecutors raided X headquarters in Paris due to concerns over cybercrime. Additionally, Elon Musk and former X CEO Linda Yaccarino have been summoned for a voluntary questioning which will take place in April.
Worldwide Responses to Grok Controversy
Most of the research and inspection into Grok’s problem leading up to and following widespread backlash was not undertaken by X itself, but instead outside organizations.With laws regulating AI being such a recent development, means that until lawmaking catches up, for now the company can continue to operate without much stricter oversight.
Grok was proposed as an anti-censorship alternative to biased traditional media, and it often gave journalists an automated reply that stated “Legacy Media Lies.” Even before early 2026, the AI model had landed itself in multiple other controversies over prejudiced and biased statements. For instance, in 2025 Grok was found to respond to provocative questions, such as naming Adolf Hitler as a figure who could “tackle anti-white hate,” bashing political leaders in Turkiye and Poland, and affirming the existence of a “white genocide” in South Africa.
Even though other AI platforms are better regulated, such as OpenAI, Google, and Meta bots which were found to refuse the same non-consensual image edits that Grok accepted, it is an example of how loosely companies can moderate their generative technologies without much consequence.
