KUALA LUMPUR, Jan 23, 2026, 23:13 (MYT)
- Malaysia restored access to xAI’s Grok after X implemented new safety and security measures, the regulator said
- Communications Minister Fahmi Fadzil said Malaysia will review the user threshold that exempts some platforms from licensing
- A research group estimated Grok generated about 3 million sexualised images in 11 days after an image-editing feature launched
Malaysia lifted its temporary restriction on access to Elon Musk’s Grok chatbot on Friday after X put in place additional safety measures, the country’s communications regulator said. (Reuters)
The decision lands in the middle of a wider push by governments to curb “deepfakes” — fake images made with AI that can mimic real people — after Grok’s image tools were used to create sexualised pictures that regulators say can harm victims fast.
Malaysia is also looking at tightening the net for social platforms that sit below its current licensing threshold. Communications Minister Fahmi Fadzil told Parliament the government would review the cut-off, saying: “Online harm does not just cease to exist when there are fewer than eight million users.” (The Straits Times)
The Malaysian Communications and Multimedia Commission said it had lifted the restriction with immediate effect after “confirmation of the implementation of additional preventive and security measures” by the platform. It said authorities would keep monitoring compliance. (CNA)
MCMC said it lifted the ban after receiving assurances at a Jan. 21 meeting with the Communications Ministry and representatives from X, The Star reported. The paper also cited an X post promising to “geoblock” — block by location — the ability to generate images of real people “in bikinis, underwear, and similar attire” where such content is illegal. (The Star)
Bloomberg reported Malaysia’s move followed discussions with the company behind Grok and came alongside ongoing monitoring by authorities, after safety measures were added to reduce the generation of sexualised images. (Bloomberg)
Malaysia imposed the restriction on Jan. 11 after what the regulator described as repeated misuse of Grok to generate obscene and non-consensual manipulated images, according to an earlier report by Channel News Asia. That report said the AI tool is integrated into X, formerly Twitter, and that Indonesia moved a day earlier to deny access to the tool. (CNA)
Last week, Malaysia’s regulator said it would take legal action against X over Grok-related safety concerns, accusing the firms of failing to remove harmful content despite notices. xAI replied to Reuters with the words: “Legacy Media Lies.” (Reuters)
Pressure on Grok has been driven in part by new research on the scale of the images. The Center for Countering Digital Hate said on Jan. 22 it estimated Grok generated about 3,002,712 sexualised images in 11 days, including roughly 23,338 images it classified as likely depicting children, based on a sample of 20,000 images from a wider pool it said totalled about 4.6 million. (Counterhate)
Imran Ahmed, the group’s chief executive, told the Guardian that “Grok became an industrial-scale machine for the production of sexual abuse material,” calling the findings “clear and disturbing.” (Theguardian)
But the lifting of the restriction does not close the file. Malaysia has signalled it will keep watching for repeat breaches, and the core risk — AI tools turning everyday photos into explicit images in minutes — has proved hard to stamp out once features spread across platforms.
For X and xAI, the Malaysia decision is a quick win on access, but it also leaves a marker: safety measures that satisfy one regulator this week may not satisfy the next one, especially as governments tighten rules around image-generating AI.
