Rhiannon Morgan AI Photo Criticism: What Happened and Why It Matters

The debate over Rhiannon Morgan’s AI-edited photo is about much more than an unflattering retouch. Morgan, a disability and visible-difference campaigner from Bridgend, Wales, said a photo intended for a workplace newsletter had been altered so that distinctive features of her skin were smoothed away. The controversy asks a direct question: should an AI image tool change how a real person looks when no one has asked it to change that person’s appearance?

The incident was reported by BBC Wales on October 29, 2026. Morgan described the result as upsetting; Canva, the design platform involved in the newsletter workflow, said it does not automatically retouch photos and was investigating what happened. Those accounts should not be collapsed into a finding that Canva deliberately altered her appearance. See the BBC Wales investigation.

What happened to Rhiannon Morgan’s photo?

According to the BBC report, a colleague was using Canva to arrange photographs for a work newsletter. Morgan understood the request to be about layout rather than altering the people pictured. But the resulting image changed her skin’s color and texture: the pinkness she normally has and hardened skin visible on her raised hands appeared to have been removed or smoothed.

Morgan, who has epidermolytic ichthyosis, saw the image before the newsletter was distributed. That distinction matters: this is not a documented case of the altered photo being publicly circulated in the finished newsletter.

Epidermolytic ichthyosis is a rare inherited skin condition associated with fragile skin, blistering, and other difficulties. Morgan previously described her experiences with healthcare, bullying, and public reactions in the Rhiannon Morgan’s account published by Changing Faces. Her criticism of the image therefore concerns the erasure of features that are part of her identity and lived experience, not simply a preference for one editing style over another.

What is confirmed, and what is still unclear?

This distinction between a person’s account, a company response, and a completed technical investigation is essential. It is reasonable to discuss the harm Morgan described without claiming that a particular Canva feature was conclusively responsible.

Why did the incident draw criticism?

1. It treats visible differences as flaws to erase

Photographic adjustments can correct lighting, cropping, or exposure. But an edit that removes scars, birthmarks, or the appearance of a skin condition can alter what an image communicates about a person. The implied standard is especially troubling when the software is asked to make someone look “normal,” as though a visible difference disqualifies the original image from being acceptable.

2. It raises questions about consent and control

Agreeing to appear in a newsletter does not necessarily mean agreeing to have distinctive physical traits changed. In workplace communications, the person photographed may have little visibility into the editing process. The practical question is who approves the final likeness, especially when AI tools are used for routine layout or cleanup.

3. Bias can enter through prompts and defaults

BBC Wales separately tested image tools using prompts about making people look “normal” and reported that Grok, Gemini, and ChatGPT removed visible differences in its examples. These were reported observations from a particular test, not evidence that every image request or product version behaves the same way. The results nevertheless illustrate how an ambiguous prompt can encode a narrow assumption about appearance.

Canva’s own Canva’s guidance on safe and lawful use of Magic Studio tells users to get permission when editing photos of recognizable people and to review outputs for unexpected or inappropriate results. That guidance describes responsible use; it does not settle the cause of Morgan’s particular experience.

How does this connect with the wider AI photo debate?

The BBC investigation also discussed people with other visible differences, including campaigner Jono Lancaster, whose appearance has been changed in AI imagery. Campaigner Katie Piper has questioned the use of AI to “fix” faces. Taken together, these examples have shifted the conversation from how convincing a generated image looks to whether the subject is represented accurately and respectfully.

There is also a data-protection dimension. In February 2026, the UK’s Information Commissioner’s Office joined international regulators in warning about realistic AI depictions of identifiable people made without their knowledge or consent. That UK Information Commissioner’s Office statement on AI-generated imagery addresses a wider range of risks, not a legal ruling on Morgan’s newsletter photo. It reinforces the importance of transparency, safeguards, and meaningful ways for people to challenge harmful uses of their image.

A practical checklist before using AI on someone’s photo

For employers, designers, communications teams, and anyone publishing real people’s photographs, the following checks can reduce avoidable misrepresentation:

• Define the edit. Specify layout, crop, or exposure changes rather than broad instructions such as “improve the person” or “make them normal.”
• Ask permission. Check whether the subject agrees to AI-assisted changes to their appearance, particularly changes affecting skin, scars, disability-related features, or facial structure.
• Preserve an original. Keep the unedited file so changes can be identified and reversed.
• Compare before and after. Inspect hands, skin texture, birthmarks, scars, facial characteristics, and other distinguishing details—not just the background.
• Use human approval. When possible, show the person the finished image before publishing it, and provide a straightforward correction process.
• Disclose meaningful alterations. Tell readers when a representation of a real person has been substantially changed, rather than presenting it as an ordinary photograph.

What is the takeaway?

Rhiannon Morgan’s criticism is not an argument against every kind of AI photo editing. It challenges an assumption that removing a visible difference is an improvement. The unresolved technical details of the newsletter image deserve careful investigation, while the broader editorial principle is already clear: tools should serve the person represented, not silently replace that person’s appearance with a standardized ideal.