X pledges safeguards in response to Grok safety concerns, says Fahmi

Social media platform X has confirmed to the Malaysian government that its AI chatbot Grok can no longer be misused to generate or edit harmful content involving pornography, sexual material or incest.

According to a report by Bernama, this follows tighter regulatory controls on its image editing and video generation features. The news portal quoted Communications Minister Fahmi Fadzil as saying, that the confirmation was conveyed to him by representatives of X for the Asia-Pacific region during an hour-long meeting at Menara Komunikasi on 21 January yesterday (Wednesday). 

Don't miss: IMDA in talks with X as Grok misuse sparks safety concerns

He said that X's management team had expressed a strong commitment to ensure that the platform becomes safer, and to work in closer cooperation with the Communications Ministry and the Malaysian Communications and Multimedia Commission (MCMC). Fahmi also shared that current takedown requests involving content on X stood at 50% as of 21 January, considered to be a moderate figure, that the minister views as a positive development. 

Based on a Malay Mail report, the MCMC received 17 complaints regarding Grok on X, consisting of six official complaints to the commission, two police reports, eight complaints through social media tagging, and one First Information Report (FIR) by the MCMC. 

Fahmi also shared in Parliament that the MCMC will verify the measures taken by X, to evaluate its effectiveness and to take follow-up action if necessary. With regards to further steps taken to safeguard social media users, Fahmi said his ministry is considering revising the eight million user threshold, in which social media platforms in the country who pass the threshold, will need to register for an Application Service Provider (ASP) licence. 

At the moment, X is exempted from the ASP licence as it has less eight million users in Malaysia. 

Indonesia was the first country to block Grok AI over the circulation of sexualised images. Indonesia’s communications and digital ministry said last week that preliminary findings showed Grok currently lacks clear safeguards to prevent the creation and spread of pornographic material using real photos of Indonesian citizens. Singapore’s Infocomm Media Development Authority (IMDA) is also engaging X over concerns related to Grok.

Last week, the MCMC blocked access to Grok for users in Malaysia, with formal notices issued to X and xAI. xAI is Elon Musk’s artificial intelligence company, which acquired social media platform X in March last year. Access to Grok will remain restricted until effective safeguards are implemented to prevent content involving women and children.

However, soon after the ban, The Guardian released a report, which claimed that Grok remained accessible to users in Malaysia despite the ban. Grok itself acknowledged that Malaysia’s DNS block was “pretty lightweight” and easy to bypass via VPNs or minor tweaks. The tool operates across multiple platforms, including a standalone app, website, and integration with X, making restrictions difficult to enforce. Users and even Grok have shared ways to circumvent the block, highlighting the limitations of such government-imposed restrictions.

A+M has reached out to the MCMC for a statement in response to The Guardian's report.

On 14 January, X announced new measures to prevent its AI chatbot Grok from generating sexualised images of real people, following global backlash against the company. Its safety team said in a statement that technological measures have been implemented to stop Grok from allowing the editing of images of real people into revealing clothing. “This restriction applies to all users, including paid subscribers,” the statement said.

Beyond Malaysia and Indonesia, regulatory pressure continues to build across the region. Hong Kong’s privacy watchdog said it is engaging relevant organisations to understand the risks posed by Grok, while reminding the public of data protection obligations when using AI tools. The Philippines also ordered the takedown of Grok, citing failures in content regulation and the spread of illicit deepfakes involving women and minors.

Related articles: 
Fahmi Fadzil calls out Meta and X for falling short on curbing online harm
Philippines orders takedown of AI chatbot Grok, following Indonesia and Malaysia 
MCMC seeks meeting with X as it prepares legal action over Grok