Grok’s bikini trend has drawn sharp criticism from Elon Musk’s company. It also raises a broader question: How do other AI chat platforms implement basic guardrails to keep online spaces safe?

India Today’s OSINT team tested major AI platforms for this very signal. While Grok emerged as the infamous one among the group, other platforms like ChatGPT, Gemini, Meta AI, and Microsoft Copilot showed more restraint. For similar indications of framing photos of women in bikinis, each cites their own safety policies regarding sexual exploitation and non-consensual content.

However, Grok’s new image-editing feature not only violates the modesty of women in the online ecosystem, but also violates at least three provisions of xAI’s own Acceptable Use Policy and Terms of Service.

Elon Musk’s AI company, XAI, launched a new image-editing feature on Grok in the last week of December. In Grok’s own words, its purpose was to allow users to make “quick, fun edits” to images. Shortly thereafter, it began to be used for more sinister purposes.

Users have widely used this feature to inspire AI to remove clothes from real women, children and celebrities.

XAI’s Acceptable Use Policy expressly prohibits the use of its services to violate an individual’s rights of privacy or publicity, to depict individuals in an obscene manner, or to sexually abuse or exploit children. Yet the image-editing feature continues to be used in ways that violate each of these restrictions.

Experts say that civil law principles do not disappear in the digital world. “A woman’s right to privacy is inherent to her dignity, and if legal action is taken, X may be held liable to compensate for the loss of dignity,” says Delhi-based cyber and arbitration lawyer Sanad Dobwal.

Despite the company’s stated ban on sexual exploitation or abuse of children, an India Today review found that such practices remain largely unregulated. When prompted with an image of minors at a summer pool party, Tool complied. This clearly violates Section 1.4 of the Acceptable Use Policy (AUP) regarding child safety.

For a long time, experts have argued that the use of AI without adequate censorship will make digital spaces unsafe. This is one such example. Even when India Today asked Grok to convert images of a real woman into “highly transparent, wet-looking bikini” scenes for testing purposes, the algorithm complied. It retained the woman’s original likeness while creating a non-consensual, sexualized image.

Dobwal mentions that “Indian Justice Code (BNS) is a law which was implemented in 2023 mainly in 3 sections, section 75, section 79 and section 356 for defamation. If someone is creating a picture, it will be the company’s fault, basically the platform’s fault”

Grok doesn’t just follow the user’s instructions, but plays with them. When asked to recreate a photo of a woman in a high-transparency, wet-looking bikini, the system did more than just create an image. It suggested variations such as “neon color” or “holographic sheen” bikinis to try further.

This behavior is inconsistent with xAI’s claims of “respecting guardrails” and keeping the service “safe for humanity.”

The tech giant has faced a lot of backlash from the public and lawmakers in India for effectively selling women’s modesty as a feature. In doing so, it has also exposed a deep uneasiness in public behavior: women who post funny vacation bikini photos are judged and labeled “sluts,” while altered images, generated without consent and reimagined by an algorithm, circulate freely as entertainment.

Rajya Sabha MP Priyanka Chaturvedi addressed the issue in a letter to IT Minister Ashwini Vaishnav, describing it as a “new trend” of fake accounts systematically targeting women.

In response, on January 2, the Ministry of Electronics and Information Technology (MeitY) issued a formal notice to X, urging the platform to conduct a comprehensive audit of Grok’s security measures and content protection measures. The directive also bans the creation of content containing “nudity, sexuality, sexually explicit, or unlawful” content.

More than 72 hours after government intervention, Grok’s new image-editing feature continues to strip women of their clothes.

Experts told India Today there are immediate, practical steps women facing AI-generated abuse can take to protect themselves. The most important step is to preserve evidence by saving screenshots or links of AI-generated images and reporting the content to both the platform and the cyber crime cell. Even if an image is fake, existing laws allow action against both the individuals creating such content and the platforms hosting it. Overall, these steps can help protect a woman’s right to privacy.

regulatory action

Australia’s online security watchdog eSafety said it had received multiple reports of Grok being used to generate sexually explicit images since late December. An eSafety spokesperson said, “Some reports relate to images of adults, which are assessed under our image-based abuse scheme, while others relate to potential child sexual exploitation material, which are assessed under our illegal and restricted content scheme.”

The European Commission said on Monday that images of naked women and children circulating on Elon Musk’s social media platform X were “illegal and appalling”.

Addressing journalists, Commission spokesman Thomas Regnier said the content violated European law. “It’s illegal. It’s appalling. It’s disgusting. That’s how we see it and it has no place in Europe,” he said.