Grok has faced a global backlash after it emerged that its image creation feature allowed users to sexualise pictures of women and children using simple text prompts.

Malaysia and the Philippines followed Indonesia in temporarily blocking access to the tool, which is integrated into X, over concerns about AI-generated sexual deepfakes.

Alexander Sabar, a senior official in Indonesia's communication and digital ministry, said access was being restored "conditionally" after X Corp gave "a written commitment containing concrete steps for service improvement and prevention of misuse".

Sabar said in a statement the ministry would continue to supervise and evaluate Grok and that it would take "corrective actions", including another suspension, if violations were found.

xAI, the Musk-owned startup that developed the AI tool, did not respond immediately to AFP's request for comment.

However, it told the Philippines last month that it would modify Grok to suit the local market, including the "total exclusion of pornographic content, particularly child sexual abuse material".

Malaysia also restored access to Grok after receiving similar promises from X that included "additional preventive and security measures".

The EU said in late January it had opened an investigation into Grok's sexualised deepfake images of women and minors.

Grok said in response it was restricting image generation and editing to paying subscribers.