Grok has made it easy for users to generate on-demand deepfake pornography and child sexual abuse material based on real people. This functionality has led to targeted harassment against women, disproportionately harming women of color. It also allowed strangers to “mock and strip” women in religious and cultural clothing like hijabs and saris, then share the images with followers on X.

Grok has aided Islamophobes, racists, and misogynists in creating this new wave of online harassment: nonconsensual sexualized and degrading imagery.

The people behind generative AI tools, like the engineers at xAI, make choices that can shape and constrain its outputs. Other AI tools have taken steps to limit the generation of pornographic images, so Grok can too.