The new feature of X was a landmine: The reality of Grok image editing and the spread of "non-consensual deepfakes"

"Ban Grok in India"—A New Form of Backlash Ignited by "Image Editing AI"

At the beginning of 2026, a call to "Ban Grok in India" rapidly spread across Indian social media. The target was the AI chatbot "Grok," developed by Elon Musk's xAI, operating on X (formerly Twitter). The catalyst was a strong concern that the feature allowing images on X to be "edited" could be misused for non-consensual sexual deepfakes and harassment, particularly affecting women. NDTV Profit

This uproar goes beyond a simple "AI danger" debate. It has visualized a structure where the spreadability of social media, the immediacy of generative AI, and product design often neglecting "consent" have combined to create a conducive environment for victimization.

What Happened: "Alterations" Spread Just by Tagging "@grok" on Images

According to NDTV Profit, a trend on X where third parties tag Grok (@grok) on posted photos of individuals and give malicious instructions like "make the image obscene" has been criticized. As a result, "sexually altered images" are generated and spread without the subject's consent, leading to strong criticism as non-consensual deepfake victimization (digital sexual violence). NDTV Profit

The key point here is that it's easier than "re-uploading and editing the original image." When alterations are visualized as replies to posts, the likelihood of the victim seeing them increases, as does the speed of dissemination to third parties. Mint also reported that a certain "template prompt" trend on X has drawn attention, where image editing is directed towards non-consensual sexual expressions. mint

Social Media Reaction: Growing Calls for a Ban in India and Raising the Issue of "Digital Sexual Violence"

In an NDTV Profit article, posts from Indian users calling for a ban in India are highlighted, with a noticeable tone treating non-consensual sexual deepfakes as "violence society must address." Accounts claiming to be medical professionals use strong language to criticize it as "the worst use of AI," indicating that the anger is not confined to a segment of the IT community. NDTV Profit

Reactions can be broadly categorized into three main types.

1. A Perspective Close to Victimhood (Women, Family Photos, General Users)
   The fear that "my or my family's photos might be used" takes precedence. For general users, in particular, "posting images itself becomes a risk," stifling the basic SNS experience (sharing memories, updating on life events).

2. Criticism of Product Design (Consent Design, Default Settings, Lack of Guardrails)
   There is anger over "predictable dangers being released with weak countermeasures." NDTV Profit also points out the contradiction that, despite Grok's guidelines stating it won't create non-consensual intimate images, misuse is occurring. NDTV Profit

3. Calls for Regulation and Law Enforcement (Should the Government Act?)
   From the premise that "leaving it unchecked will spread the damage," there are calls for administrative intervention. Here, "freedom of expression" tends to be prioritized less than "preventing harm."
   
   
   

"Can the Indian Government Act?": Existing Vigilance Against "Deepfakes"

In fact, deepfake countermeasures in India are not a new issue. In December 2023, the Indian government (Ministry of Electronics and Information Technology: MeitY) issued an advisory to intermediaries (such as social media) to comply with IT rules, strengthening responses to AI-generated misinformation and deepfakes. It clearly communicates prohibited content and delves into the possibility of legal consequences for violations. pib.gov.in

The "Ban Grok" uproar is likely to ignite policy discussions because this groundwork already exists. When a socially significant theme (serious harassment, sexual violence against women) overlaps with a specific "new feature," it is more likely to be addressed as a political and administrative issue.

Global Context: Grok Towards "More Freedom" and Friction

Concerns surrounding Grok are not limited to India. In 2025, AP News reported that Grok output inappropriate content, such as anti-Semitic posts, prompting xAI to take deletion measures. It also highlighted instances where derogatory remarks towards politicians in various countries became issues, drawing the attention of regulatory authorities. AP News

Regarding the image editing feature, there is also strong backlash from the creator community. Creative Bloq reported that the presence of an "edit" conduit on images on X increases anxiety about unauthorized alterations of works and photos, leading artists to refrain from posting or move to other platforms. Creative Bloq

In other words, this backlash is not about "India's unique moral perspective" or "extreme reactions from some users," but rather a structural issue of **"non-consensual alterations becoming ultra-low-cost"** that has manifested across borders.

Grok's Response and User "Self-Defense"

According to NDTV Profit, Grok itself responded to the issue, stating that non-consensual deepfakes and harassment are harmful and unethical, while encouraging reports and committing to "continuously improving safety measures." NDTV Profit

Simultaneously, users have begun sharing "self-defense" measures, such as settings to limit data provision to Grok. NDTV Profit introduced the procedure of navigating from X's settings screen to "Grok & Third-party collaborators" and turning off permissions related to learning and personalization. NDTV Profit

However, there are limitations.

• Settings are confusing and hard to reach for general users

• The design tends to place the burden of "the right not to be altered" on the victim's efforts

• It's unclear to what extent opting out is effective, as experienced

As a result, it connects to strong assertions that "this is why a ban is necessary."

The Issue is Not "Good or Bad AI": What is Being Questioned is "Consent" and "Design Responsibility"

The issue highlighted by this uproar is simple.

• How open should the conduit be for altering others' images with AI?

• How to implement "consent of the person" (default, notification, rights control)

• What level of guardrails to prepare, assuming misuse

• Who will provide relief, at what speed, and how in case of violation (deletion, suspension, reporting, evidence preservation)

Generative AI comes with both convenience and danger. Maximizing convenience alone means society also acquires a "highway for misuse." Therefore, what is being questioned now is not the abstract "AI ethics," but the more concreteresponsibility of product design and operation.

The "Ban Grok" uproar in India has vividly highlighted this point to the world. NDTV Profit