🧔♂️ A friendly human may check it before it goes live. More news here
Indonesia sets conditions to remove ban on Grok AI
Indonesia’s Ministry of Communication and Digital has outlined conditions for lifting the ban on Grok AI, the chatbot integrated with X and developed by xAI.
Officials require adjustments to Grok AI’s algorithms and stronger content moderation to prevent illegal content, including non-consensual sexual deepfakes.
Platforms must also comply with Indonesia’s electronic system operator registration rules before access is restored.
The ministry said the ban is temporary, and will be lifted once Grok AI meets legal requirements, especially those related to content moderation and user protection.
Authorities began blocking access to Grok AI after reports of its misuse to create and distribute AI-generated pornographic deepfakes.
X has reportedly committed to making technical changes in line with local regulations, with a written commitment expected soon.
The government said its main concern is public safety and digital security, and it will continue monitoring all digital platforms for compliance.
🔗 Source: Bisnis Tekno
🧠 Food for thought
Implications, context, and why it matters.
Indonesia’s 4-hour takedown law creates an enforcement bottleneck that Grok must solve
- MR5 requires platforms to remove content labeled “unlawful” within 24 hours, or within 4 hours for urgent cases such as pornography 1. Grok AI needs automated systems that spot and block non-consensual deepfakes within those windows which helps explain why the ban is not easy to lift.
- Fines can reach $33,000 per violation, with penalties multiplying for repeat offenses 1. A chatbot that generates content must prove it can stop illegal outputs before they appear, not only clean them up after, so Grok needs deeper algorithm changes than routine filters.
AI companies building Indonesian deepfake detection tools can capture urgent compliance demand
- Real-world tests put deepfake detection at about 69% accuracy 2, while platforms face 4-hour deadlines for urgent removals 1. A 2025 Indonesian study measured 91.25% accuracy on a 2,000-image dataset using XceptionNet models (a family of convolutional neural networks commonly used for image classification and deepfake detection) 3, which signals growing local capability.
- Election monitors counted more than 3,200 election hoaxes since July 2023 4. Kominfo, Indonesia’s communications ministry, asked Google to delist over 500,000 URLs between June and December 2021 1. Any chat app, social network, or User-Generated Content (UGC) service entering the market needs MR5-ready moderation, so vendors with turnkey workflows can serve Grok AI along with other platforms under these timelines.
Stay updated on the go with our mobile app.
Get latest insights with smoother, more personalized experience through TIA mobile app.