X accepts fault on Grok AI obscenity issue, pledges compliance with Indian law
Microblogging site X has accepted its mistake and assured it will comply with Indian laws after a warning from the IT Ministry on AI-generated obscene content linked to Grok. According to government sources on Sunday, January 11, 2026, around 3,500 pieces of content have been blocked and over 600 accounts deleted.
Sources said X has committed that such imagery will not be allowed going forward. The platform has also agreed to provide the details the government asked for and to align future enforcement with statutory requirements.
From inadequate response to corrective commitments
After the first notice, X shared a lengthy reply outlining its strict takedown policies for misleading posts and non-consensual sexualized images. But it missed key information: specific takedown details tied to the Grok AI incident and preventive measures to stop a repeat.
On January 2, the IT Ministry issued a stern warning over indecent and sexually explicit content generated via Grok and similar AI tools. X's Safety team stated on Sunday, January 4, that it removes illegal content, permanently suspends accounts, and works with law enforcement, including on CSAM. The company added that anyone using or prompting Grok to create illegal content will face the same consequences as if they uploaded it.
Regulatory context and signals for platforms
This episode underscores a clear expectation: AI-assisted generation is subject to the same legal standards as uploads by users. Intermediaries must be able to identify, act on, and document AI-driven violations with the same rigor as traditional posts.
For reference, see the Information Technology (Intermediary Guidelines and Digital Media Ethics Code) Rules, 2021, issued by MeitY. These rules frame obligations around due diligence, takedown timelines, grievance redressal, and cooperation with authorities.
Action points for government, legal, and platform teams
- Require detailed post-incident reports: volume, categories, timestamps, detection methods, and account-level actions specific to AI-generated content.
- Mandate preventive controls: prompt filtering, blocked terms, model-side safety rails, and real-time human review for flagged generations.
- Insist on auditable logs that link prompts, generations, user IDs, and enforcement outcomes to support lawful requests.
- Set clear escalation protocols to the Grievance Officer and Nodal Contact, with documented response times and closure notes.
- Publish regular transparency updates that separate AI-generated violations from traditional uploads for better risk visibility.
- Strengthen cooperation with local law enforcement and trusted flaggers, including proactive sweeps on repeat vectors and clones.
- Run red-team tests on AI tools (like Grok) focused on sexual content, non-consensual imagery, and child safety, and show measurable reductions in successful prompts.
What to watch next
Whether X delivers the missing takedown specifics and a concrete prevention plan will set the tone for future oversight. Sustained reduction in AI-driven obscene content, repeat offender controls, and faster cycle times on removal will be the practical markers of progress.
Expect closer scrutiny of other AI features that enable content generation. Clear documentation, faster cooperation, and consistent enforcement will matter more than statements of intent.
Need to upskill your team on AI risk and compliance? For structured training built around job roles, see the curated programs at Complete AI Training: Courses by Job.
Your membership also unlocks:
