By Karim Hammam
A BBC Wales investigation has found that popular artificial intelligence tools remove scars, birthmarks and other visible differences when prompted to make a person look "normal." Testing five leading AI systems, reporters discovered that xAI's Grok, Google's Gemini and OpenAI's ChatGPT all erased facial differences on the first attempt, while leaving clothing, hair and background figures unchanged.
AI Testing Results
Anthropic's Claude Opus 5.5 refused the prompt, stating that erasing scarring changes a real person's identity without consent. Meta's AI Thinking also declined, responding that the person already looks normal. On a second test image, Gemini refused while Grok and ChatGPT continued to make the alterations. Jaymie Thomas, co-founder of AI Wales, explained that these tools rely on training data where conventional appearances dominate, making them more likely to reproduce a narrow standard.
Personal Impact
Rhiannon Morgan, 30, from Bridgend, said an image of her used in a work newsletter was auto-retouched by Canva's AI, smoothing her skin and removing the pink tone and hard skin caused by epidermolytic ichthyosis, a rare genetic condition. "It eradicates all the obstacles I've had to overcome to be the person I am today," she said. Kirsty Hill, 33, from Maesteg, who has Fraser Syndrome, described feeling "angry and violated" by the investigation's results. Jono Lancaster, a campaigner with Treacher Collins syndrome, called AI-altered images of himself "super offensive" after years of self-acceptance work. The issue gained wider attention when TV presenter Katie Piper, who survived an acid attack in 2008, posted about Grok being asked to "fix her face."
Company Responses
Canva apologized and said it does not automatically retouch photos, adding it is investigating the case. Google stated Gemini is designed to follow editing instructions with safeguards against harmful content and works to improve when systems respond inappropriately. OpenAI said reducing bias is an ongoing priority and it is committed to representing user diversity. xAI did not respond to a request for comment.
Calls for Change
Changing Faces, a charity estimating five million people in the UK have a visible difference, said it is unacceptable that people are excluded from AI or forced to conform to narrow beauty standards. Chief executive Louise Wright urged more inclusive AI development. Lancaster is working with Face Equality International to create a more diverse database for AI training.
