A BBC Wales investigation found AI image editors can erase visible differences when asked to make people look “normal”. Rhiannon Morgan said AI “eradicates all the obstacles” she overcame, highlighting consent and representation risks as models impose narrow beauty standards.
What Happened
A BBC Wales investigation revealed that AI image‑editing tools such as Elon Musk’s Grok, Google’s Gemini and OpenAI’s ChatGPT can automatically remove visible differences from photographs when prompted to make a person look “normal”.
The study focused on people with scars, birthmarks and medical conditions, a group estimated by the charity Changing Faces to number about five million in the UK.
The investigation found that the AI tools altered only the visible differences; they left clothing, hair and background people unchanged.
Rhiannon Morgan, 30, from Bridgend, said an image of her used in a work newsletter had been auto‑retouched by AI.
She described the final result as “one of those really jarring moments where I thought I don’t know who that person is.” Morgan added that the AI “eradicates all the obstacles I’ve had to overcome to be the person I am today” and that “in literally seconds, AI can eradicate the hospital appointments and bullying I’ve been through.” She also noted, “I’m quite pink usually and that had gone,” and, “I’ve also got hard skin on my hands and that had also been smoothed over.”
Canva, the platform that hosts the newsletter image, responded: “We’re sorry to hear about this experience and take matters like this very seriously. Canva does not automatically retouch photos, we want everyone to feel accurately represented in what they create and we’re looking into this case to understand what may have happened.”
The issue was first raised by campaigner Katie Piper after a user uploaded a photo of her to Grok and asked the model to “fix her face”.
Piper, who has undergone hundreds of surgeries following a 2008 attack, posted on social media: “So, today I found out what I’d look like if I was ‘normal’… Someone uploaded a photo of me to Grok and asked it to ‘fix her face’.” Her post sparked thousands of supportive comments and the question: “What happens when AI decides what ‘normal’ looks like?”
BBC Wales tested the claim by feeding the same image to the three AI tools and comparing the outputs.
All three models removed visible differences such as Morgan’s pink skin tone and hard skin on her hands, yet left other facial features, hair, clothing and background people untouched.
Changing Faces, a charity supporting people with visible differences, stated that it is “unacceptable that people are excluded from AI or are made to conform to narrow beauty standards.” The charity’s estimate of five million affected individuals underscores the scale of the problem.
What This Means For You
If you work in marketing, design or content creation, you should verify that any AI‑enhanced images retain the authentic features of subjects, especially when those subjects have visible differences.
Because the investigation shows AI can silently erase such differences, you may need to add a manual review step or use tools that flag alterations.
For individuals with visible differences, consider reviewing AI‑edited images before publication. The source story notes that Rhiannon Morgan saw the altered image before it was sent out, which allowed her to challenge the changes.
If you rely on AI for photo editing, you might want to keep a backup of the original file and compare it to the AI output.
If you are a developer of AI image‑editing tools, the BBC Wales findings suggest you should implement safeguards that prevent the automatic removal of medically or personally significant features.
The fact that the models altered only the visible differences indicates a narrow training objective; expanding the model’s understanding of diversity could mitigate this issue.
Why It Matters
This incident highlights a broader ethical dilemma: AI systems can impose a narrow definition of “normal” that erases individuality.
The BBC Wales report shows that even when users explicitly request a “normal” look, the AI interprets that as a call to remove visible differences, effectively erasing the person’s medical history and identity.
This raises questions about consent, representation and the potential for AI to reinforce harmful beauty standards.
Moreover, the fact that the AI tools left clothing, hair and background people unchanged suggests that the models are not applying a blanket smoothing filter but are targeting specific visual cues.
This targeted approach could be exploited to subtly alter public perception of individuals with visible differences, which could have implications for media representation, hiring practices and social inclusion.
For the AI industry, the BBC Wales investigation serves as a reminder that model training data and prompts can produce unintended social consequences. Companies must audit their models for bias and ensure that users have control over how personal features are treated.
Key Takeaway
- AI image‑editing tools can automatically erase visible differences when prompted to make a person look “normal”.
- Five million UK residents are affected by visible differences, making this a widespread representation issue.
- Canva and other platforms must implement safeguards and manual review processes to prevent unconsented alterations.
- Users with visible differences should review AI‑edited images before publication to protect their identity and history.
Frequently Asked Questions
What AI tools were involved in the BBC Wales investigation?
The investigation tested Elon Musk’s Grok, Google’s Gemini and OpenAI’s ChatGPT.
Did the AI tools change anything other than visible differences?
No. The tools left clothing, hair and background people unchanged.
How many people in the UK have visible differences?
The charity Changing Faces estimates about five million people are affected.
Leave a Reply