DShortly after Malaysia made global headlines by announcing that it would temporarily ban Grok over its ability to generate “highly offensive and non-consensual manipulated images”, the generic AI tool was found to be interacting seamlessly with accounts registered in the country.
“Still here! The DNS block in Malaysia is quite mild – easy to bypass with a VPN or DNS tweak,” Grok’s account on X said in response to a question from a user.
Grok’s ability to allow users to create sexually explicit images, including images of children, has sparked global outrage in recent weeks, with regulators and politicians around the world launching investigations. Indonesia and Malaysia became the first two countries to announce a block on the technology, with Malaysia’s regulatory body saying last Sunday that it has “directed a temporary ban” on access to Grok, effective from January 11, 2026. Philippine authorities have said they also plan to ban the technology.
However, blocking access to Grok is not straightforward. This technology is not only present on multiple platforms, including a standalone app and website, but is also integrated into X, which, along with Grok, is owned by Elon Musk’s XAI.
Over the past week, X users and even Grok himself have advised people to avoid restrictions. This includes using a VPN – many of which are available for free – or changing the Domain Name System (DNS), the protocol on the Internet that converts address names to IP addresses that load websites.
When Guardian tried to use Grok in Indonesia, its website was working even without a VPN, although the Grok app was not working. Grok was still responding to Indonesian accounts on X, where it acts as an integrated chatbot. X is not banned.
Nana Nwachukwu, an AI governance expert and PhD researcher at Trinity College Dublin, said even if governments could ban Grok entirely, that’s still not a real solution.
He said, “Stopping the grok is like slapping a Band-Aid on a weeping wound that you haven’t cleaned.” “You block Grok, and then you keep screaming that you did something. Meanwhile, people can use VPNs to access those same platforms.” Or, they could simply turn to one of the many other platforms that provide similar functions, including “smaller, general-purpose AI systems that are largely unknown,” Nwachukwu said.
He said governments should instead focus on law enforcement and investigating individuals who use such devices to break the law. “Platforms are required by law to provide information to law enforcement when a crime has been committed,” Nwachukwu said. “If we see that people are being arrested for these crimes, people are being prosecuted in courts, people are being sent to jail, then that is an indication that this is a real crime,”
एक्स को अपने प्लेटफ़ॉर्म में जवाबदेही बनानी चाहिए – और खुद को साफ़ करना चाहिए, नवाचुकु ने कहा: “उन सभी अपमानजनक छवियों को प्लेटफ़ॉर्म से हटा दिया जाना चाहिए।”
On Wednesday, However, the Guardian found that it is possible to get around such restrictions by using a standalone version of Grok, which is easily accessible via a web browser, to create short videos in which clothes are removed from images of real women.
It can then be posted to X’s public forum, where it can be viewed by users around the world in a matter of seconds.
Musk’s company also said that, in jurisdictions where such content is illegal, it would geoblock the ability of all users in those locations to generate images of real people in bikinis or similar attire in Grok on X, adding that “XAI is implementing similar geoblocking measures for the Grok app”.
Experts warn that users can still bypass such “geoblocks” through VPNs. It is also not clear in which countries such restrictions will be implemented.
According to local media reports, Fahmy Fadzil, the communications minister in Malaysia, has said that the ban on Grok will only be lifted when its ability to produce harmful content is disabled.
Dr. Nuuriyanti Jalli, visiting fellow in the Media, Technology and Society Program at ISEAS – Yusof Ishak Institute, said the threat of blocking Grok could be a useful way to put pressure on companies to respond quickly, adding that it “shifts the debate from ‘individual bad actors’ to questions of platform responsibility, security by design, and accountability when security measures fail.” This can “slow the spread of abuse, reduce casual abuse and create a clear boundary around what authorities consider unacceptable.”
In Indonesia, Grok has been used to create non-consensual sexual images of singers and celebrities, including JKT48, one of the country’s most popular girl groups, while women in Malaysia report similar abuse, including cases where the device has been used to remove their hijab, according to Malaysian media.
Some women resorted to publicly telling Grok on X that they did not authorize it to “crawl, take, process or edit” any of their photos.
Nuurianti said governments should “push for greater transparency about how safeguards are implemented, how reports of abuse are handled and what enforcement steps are taken when harmful content is generated or disseminated.”
The Malaysian Communications and Multimedia Commission and the Communications Ministry did not respond to requests for comment. Indonesia’s Ministry of Communications and Digital Affairs also did not respond.
Nwachukwu said safeguards should be built into the AI system, not “gates” built around it. “Both (geographical) restrictions from X, (and) restrictions from the government are gated access, and the gates can be broken,” he said.
Additional reporting from Hidayatullah
