Lead: On Wednesday, 14 January 2026, Elon Musk said he was not aware that xAI’s generative model Grok had produced any “naked underage images,” asserting on X that there were “literally zero” such outputs. His statement comes as regulators, lawmakers and rights groups escalate scrutiny of Grok and X across multiple countries. Calls are growing for Apple and Google to remove X from their app stores, the UK regulator Ofcom has opened an investigation, and several nations including Malaysia and Indonesia have restricted access or pursued legal action. X last week limited Grok’s publicly available image-generation features for many users amid those pressures.
Key takeaways
- Elon Musk said on 14 January 2026 he was unaware of any naked images of minors produced by Grok and posted that there were “literally zero.”
- Three Democratic US senators asked Apple and Google to remove X and Grok from their app stores, citing the spread of non-consensual sexual images of women and minors.
- Ofcom has launched an investigation into Grok as the UK prepares a law this week that would criminalize creating such images.
- Malaysia and Indonesia have blocked access to Grok and are pursuing legal measures against X and xAI for alleged failures to prevent harmful content.
- X curtailed Grok’s public image-generation and editing features for many users last week, though experts say safeguards may not fully prevent misuse.
- Musk emphasized Grok was designed to refuse illegal prompts and that the tool does not generate images unless prompted by users.
Background
The controversy revolves around Grok, an AI assistant developed by xAI and integrated into the X platform, which can produce text and images on user request. As generative models gained popularity, platforms and regulators have grappled with the risk that bad actors can prompt systems to create sexual content involving adults or minors without consent. In recent months watchdogs and advocacy groups have reported instances where models produced sexually explicit images or where such tools were used to alter or create non-consensual imagery of real people. Governments are moving to tighten rules: the UK is introducing legislation to criminalize creating sexual images of minors produced by AI, while other states are exploring restrictions on how generative models operate and are distributed.
Platform operators face a difficult technical and legal environment: models respond to prompts from users and may surface harmful outputs unless guarded by layered safety measures. Historically, content moderation relied on a mix of automated filters, human review and terms-of-service enforcement—methods that have been strained by the speed and scale of generative AI. Companies like xAI and X argue that policy, engineering controls and user enforcement can reduce abuse; critics contend those measures often lag behind new ways the tools are misused. International responses have begun to diverge, with some countries blocking access quickly while others pursue regulation and enforcement through courts or telecom regulators.
Main event
On 14 January 2026 Musk posted on X that he was “not aware of any naked underage images generated by Grok” and that the count was “literally zero.” He reiterated that Grok was programmed to refuse illegal requests and that it only produces images when prompted by users. Musk also said anyone using Grok to create illegal material would face consequences comparable to those for uploading illegal content themselves. His remarks arrived amid escalating public pressure: lawmakers, rights groups and platform watchdogs pressed Apple and Google to remove X from their app stores until Grok’s risks were addressed.
Last week X limited Grok’s ability to create or edit images publicly for many users, moving some capabilities behind restrictions and access controls. Industry specialists and watchdog groups, however, reported that the model retained the technical capacity to produce sexually explicit imagery under certain prompts or via alternate access paths. Those experts warned that measures like paywalls or partial feature rollbacks may reduce casual misuse but not fully block determined abusers with technical know-how or paid access.
Internationally, Malaysia and Indonesia have already blocked user access to Grok and initiated or signaled legal actions against X and xAI, saying the companies failed adequately to prevent harmful content and protect users. In the UK, Ofcom has opened an investigation and the prime minister, Keir Starmer, said on Wednesday that X was working to comply with the incoming rules criminalizing such image creation. In the United States, three Democratic senators formally urged Apple and Google to remove X and Grok from their respective app stores pending fixes.
Analysis & implications
The incident highlights a structural tension in generative AI: models are trained to respond to user prompts, which makes them powerful but also creates avenues for abuse when governance is incomplete. Even when companies build refusal behaviors and content filters, adversarial prompting and model fine-tuning can sometimes circumvent those defenses. That means legal and policy steps—criminalizing certain outputs, imposing platform liability or requiring technical audits—are likely to become central to how governments manage risks from image-generating AIs.
For platforms, the episode demonstrates cascading business risks. App-store removals or regulatory blocks in large markets can reduce user reach and revenue while causing reputational damage. Compliance with divergent national laws increases operational complexity: a setting that satisfies one jurisdiction’s rules may violate another’s. Companies may be forced to regionalize models and enforcement stacks, adding cost and fragmenting user experience.
Technically, preventing illicit outputs requires more than simple keyword blocks. Effective mitigation typically involves layered approaches: safety-aligned model training, real-time moderation, watermarking or provenance tracking, robust user authentication and cooperation with law enforcement. Even then, balancing legitimate user capabilities against safety constraints remains an open engineering and policy challenge, and no single measure is likely to eliminate misuse entirely.
Comparison & data
| Jurisdiction | Action to date | Notes |
|---|---|---|
| United Kingdom | Ofcom investigation; new law criminalizing creation | PM said X is working to comply (14 Jan 2026) |
| Malaysia | Access blocked; legal action signaled | Regulator-level restrictions in place |
| Indonesia | Access blocked; legal action signaled | National authorities pursuing remedies |
| United States | Senators requested app store removals | Legislative pressure and congressional oversight |
The table summarizes actions described publicly as of 14 January 2026. National responses range from regulator investigations to outright access blocks and legal proceedings—showing a mix of preventive, punitive and oversight measures. The variance suggests companies operating generative tools must prepare for fragmented legal regimes and fast-moving enforcement. Data-driven monitoring—reporting volumes of flagged prompts or blocked outputs—will be needed by regulators to assess whether mitigations work.
Reactions & quotes
Officials and advocates reacted quickly after reports of problematic outputs and Musk’s statement.
“I not aware of any naked underage images generated by Grok. Literally zero.”
Elon Musk (X post, 14 Jan 2026)
This short post from Musk asserted no known incidents and emphasized the platform’s built-in refusal behavior; it did not, however, include public evidence or an audit of outputs.
“X is working to comply with the new rules.”
Keir Starmer, Prime Minister (statement cited 14 Jan 2026)
The prime minister’s comment accompanied the UK’s legal changes and Ofcom’s inquiry, stressing expectations that platforms adapt to newly criminalized conduct.
“Partial restrictions and paywalls are unlikely to stop determined abusers or close all access routes to harmful image generation.”
Independent watchdogs and experts (public statements)
Experts noted that engineering controls can reduce risk but not fully prevent misuse without comprehensive systems and external oversight.
Unconfirmed
- Whether any explicit images of minors were actually produced and circulated via Grok remains unverified by an independent audit.
- It is not confirmed how many users, if any, successfully used Grok to create illegal content before recent restrictions.
- The effectiveness of X’s recent curtailing of public image features in preventing determined misuse is not yet proven.
Bottom line
The dispute over Grok underscores a growing governance gap for generative AI: companies may assert safety-by-design, but regulators and rights groups demand verifiable evidence and enforceable safeguards. Legal steps—like the UK’s new criminal provision—and regulator scrutiny will force platforms to demonstrate technical fixes, transparency and cross-border compliance. For X and xAI, the immediate priorities will be rapid audits, clearer technical mitigations, and cooperation with regulators to avoid app-store removals and further national blocks.
Longer term, policymakers and industry must develop standards for model safety, auditing and provenance that work across jurisdictions. Absent credible, demonstrable controls, companies risk sustained legal challenges, market restrictions and loss of user trust; at the same time, overly blunt restrictions could stifle legitimate innovation and user utility. Observers should watch for independent audits, published incident tallies and the responses of major app stores in the coming weeks.