Key facts
- AI tools from Google, OpenAI, and Elon Musk's Grok have been found to alter or erase visible differences in people's images.
- Some AI models, like Anthropic's Claude Opus 5.5, refused prompts to alter faces, citing ethical concerns and consent.
- Individuals with scars, birthmarks, and medical conditions reported feeling violated and that AI changes diminish their life experiences.
- The charity Changing Faces estimates five million people in the UK have a visible difference and calls AI's narrow beauty standards unacceptable.
AI tools are facing criticism for altering or erasing visible differences in people's images, with individuals reporting that the technology enforces narrow beauty standards. A BBC Wales investigation found that AI models from Google, OpenAI, and Grok changed features like scars and birthmarks when prompted to make a person look 'normal'. Some AI systems, however, refused these requests, citing ethical concerns. People with visible differences expressed anger and distress, feeling that these alterations diminish their identities and the challenges they have overcome.