{
  "id": 10450420,
  "title": "Humans Are Reading Copilot Prompts — And They're Horrified",
  "url": "https://urgent.news/2026/09/28/humans-are-reading-copilot-prompts-and-theyre-horrified",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-28T13:26:31.000Z",
  "source": {
    "name": "404 Media",
    "slug": "404-media",
    "url": "https://www.404media.co/humans-reading-copilot-prompts-images/"
  },
  "original_language": "en",
  "account": "Human contractors employed by Microsoft to enhance its Copilot AI chatbot are frequently confronted with lewd or sexually explicit image editing requests, along with graphic images uploaded by users, including upskirt photos and women in sexual positions. These contractors are tasked with evaluating whether the generated image meets the prompt's requirements, such as enlarging a woman's AI-enhanced breasts. Users uploading a picture of themselves or someone else likely assume the image remains private between them and the AI tool. However, a workforce of hundreds of human reviewers is sometimes confronted with these prompts and images. Many contractors are inundated with requests for creating foot fetish images of children's cartoon characters, shortening a real woman's skirt, or placing people in sexual positions. The internal contractor documents, revealed by 404 Media, demonstrate that AI companies are utilizing human workers to review both AI chatbot users' text prompts and the pictures they upload for editing purposes. Contractors working on Copilot have expressed horror at reviewing such explicit content, including upskirt photos and sexual positions. One contractor, speaking anonymously, described seeing pro-anorexia content and prompts designed to create lewd images of young girls.",
  "summary": "Human contractors are reviewing Copilot users’ prompts and uploaded images, according to internal documents obtained by 404 Media. The contractors are also bombarded with users’ requests for sexual AI images.",
  "key_points": [
    "Contractors horrified by explicit content in AI editing requests",
    "Upkirts, sexual positions, and pro-anorexia prompts encountered",
    "Human reviewers tasked with evaluating AI-generated image compliance"
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}