Elon Musk’s main artificial intelligence chatbot, Grok, has been found creating sexualized images of people even when users clearly warn that the subjects do not agree. This was reported by Reuters.
Earlier, Musk’s social media company X announced new limits on what Grok can post publicly. To test the chatbot, nine Reuters reporters gave it different prompts to see if it would still generate sexualized images without consent.
The investigation found that while Grok’s public X account no longer produces as many sexualized images, the Grok chatbot itself still creates such images when asked. This happens even when it is warned that the people in the images are vulnerable or could be embarrassed or humiliated.
The findings raise concerns about the chatbot’s ability to respect consent and safety, despite efforts to restrict harmful outputs in its public posts.