Microsoft Copilot contractors report reviewing sexualized, sometimes CSAM-adjacent, image-edit requests
2026-09-28
Per 404 Media, internal documents show Microsoft hires human contractors to review Copilot users' photo-editing prompts and uploaded images, including requests to sexualize real photos, AI-enlarge body parts, shorten skirts, or generate sexualized images of children's cartoon characters, plus pro-anorexia content (one prompt reportedly asked for an image of Ariana Grande with anorexia). Contractors say they are tasked with judging which of two edited images looks better, not with flagging or vetting harmful content. 404 Media notes this parallels its earlier reporting that OpenAI has thousands of contractors reviewing real ChatGPT prompts.
Significance 4: Investigative reporting backed by internal documents showing a major AI vendor's human-review pipeline is structured around judging sexualized/exploitative content quality rather than flagging harm is a serious operational/safety accountability failure, capped below 5 because it is single-sourced to one outlet with anonymous contractors and no named vendor confirmation.
Implications · machine-drafted, not owner judgment
This extends a pattern 404 Media already documented at OpenAI to Microsoft's Copilot, suggesting that routing sexualized or exploitative user content through underpaid contractors for quality judgment — rather than harm triage — may be a broader vendor practice rather than a single company's lapse. It is a direct instance of the technology-illusion pattern: a safety-review function exists on paper, but no one is accountable for the actual harm-prevention outcome. The CSAM-adjacent content described also raises real regulatory and law-enforcement exposure that could move faster than any one vendor's internal fix.
- Regulatory or law-enforcement response citing this reporting
- Whether Microsoft changes its contractor review process or scope
- Whether a similar disclosure surfaces at Google, Anthropic, or xAI's image tools
Sources
- press Humans Are Reading Copilot Prompts — And They're Horrified retrieved 2026-09-28