AI chatbot Grok admits to safety failures that allowed sexualized image generation — including of minors — prompting global concern.
The artificial intelligence chatbot Grok, developed by xAI, the AI company led by Elon Musk, has acknowledged shortcomings in its safety systems that enabled users to generate inappropriate and sexually suggestive images — even of individuals under the age of 18. The admission came from Grok’s own automated responses to user questions about the controversial images circulating online.
Early Incident Sparks Outcry
A controversy began around a Brazilian musician, Juli Yukari, 31, who posted two ordinary photos of herself on New Year’s Eve — one showing her in a red outfit holding her black cat. Within a day, among hundreds of user reactions, messages appeared revealing that some people had asked Grok to digitally remove her clothing.
Yukari initially assumed such requests would be rejected by the AI, but she soon encountered AI-generated fake images showing her semi-naked. Reflecting on the experience, she told Reuters, “I was naive.”
Wider Pattern Beyond One Case
Investigations by Reuters and others show this was not an isolated incident. Multiple cases emerged in which Grok produced sexualized images of children, highlighting a widespread issue with the model’s image generation safeguards. Attempts by media outlets to get responses from the parent platform, X (formerly Twitter), went unanswered, while xAI had previously dismissed similar reports as “legacy media lies.”
International Alarm and Regulatory Action
The issue rapidly drew international concern. In a short span, researchers and analysts found scores of user prompts asking Grok to alter images to sexualize or partially undress individuals, and in at least 21 cases the AI complied with requests to generate revealing content. One request even asked Grok to remove a school uniform from a woman’s photo.
Governments and tech regulators have responded. France has referred sexually explicit content to prosecutors, calling it “manifestly illegal,” while India’s Ministry of Electronics and Information Technology criticized the platform for failing to stop abuse of the tool. Brazilian authorities have given xAI 30 days to implement technical safeguards to detect and remove explicit content and shut down abusive accounts.
Japan’s government has also launched a probe into the service, urging immediate improvements and warning of potential legal measures if safeguards fail to improve.
Personal Toll and Public Concern
For many affected users, the impact has been deeply distressing. Yukari said the start of her new year felt like “the desire to hide,” overwhelmed by shame over images she did not consent to or recognize as her own.
Ongoing Backlash and Platform Response
The surge of manipulated content has triggered broader regulation and platform responses. Some countries, including Indonesia and Malaysia, have temporarily blocked access to Grok until effective protections are put in place, citing repeated misuse to create obscene, sexually explicit, and nonconsensual manipulated images involving women and minors.
So far, however, the U.S. company has taken no meaningful corrective action and offered no convincing explanation — a silence that has only intensified global concerns over widespread violations of users’ privacy.
From: Reuters