Hundreds of human contractors are sifting through user prompts for Microsoft’s Copilot AI, encountering a disturbing mix of dark and explicit content. A report from _404 Media_ reveals that these contractors evaluate prompts related to image editing, often facing unsettling requests. Many of these prompts involve upskirt photos and other sexual scenarios, raising serious ethical concerns about the content produced by AI.
The contractors primarily assess how well the AI fulfills these prompts. Microsoft has instructed them to use their “intuition” during this evaluation. This means they are not judging the appropriateness of the requests but simply determining if the AI meets user needs, such as enlarging a woman’s breasts or modifying a character’s clothing to be more revealing.
Some of the requests they review are particularly alarming. Some requests involve creating sexual scenarios with children’s cartoon characters, raising serious ethical concerns. Additionally, the images they evaluate often feature uncensored faces, which adds another layer of concern given the sexual nature of many prompts. “Faces are always uncensored, and many of the prompts are sexual in nature and dubiously consensual,” said one contractor who wished to remain anonymous.
Many contractors work for low wages while viewing deeply disturbing and even traumatizing content. They have expressed shock at the nature of the requests they are expected to handle. One worker described their reaction to encountering pro-anorexia content, noting that there was no prior warning before being exposed to such disturbing material. Another contractor recounted coming across a set of eight upskirt photos and asked superiors to flag the content as unsafe.
This situation reveals a hidden human aspect in the operation of AI technologies. Users often assume their interactions with AI are private and monitored only for moderation. However, human contractors are sifting through potentially problematic content. This raises significant privacy and ethical concerns about what details may be shared with external workers.
As contractors navigate their experiences, some have begun to question the nature of their roles. They wonder who is behind these prompts and why there seems to be a focus on generating pornographic content. “It’s hard to take things seriously when my focus has to be what model generated the appropriate bust size or which middle-aged woman was put in the appropriate sexually suggestive position,” remarked one contractor.
The process of evaluating user prompts for AI reveals a troubling reality about the content that is being generated and the human cost involved. Contractors face not only the challenge of managing disturbing requests but also the ethical implications of their work, as they navigate a landscape where user privacy may be compromised. The focus on fulfilling potentially harmful prompts raises questions about the broader responsibilities of companies like Microsoft in overseeing the content produced by their AI tools.
