Mia Zelu is the Instagram influencer @Miazelu. She spent the Wimbledon fortnight posting images of herself at the tennis tournament to her 168,000 followers. In the stands watching a match, in a pink tennis-like outfit, drinking a spritz. She asks fans which match they enjoyed the most. Thousands of people congratulated her on getting prime seats at Wimbledon. There is one big problem. Mia Zelu does not exist. She is AI-generated.
The creator of Mia Zelu has intentionally not disclosed their name or affiliations. The account’s bio lists Mia as a “digital creator”. Clear as mud!
You may scoff that her followers thought she was real because of the high quality of AI. You may go further and laugh at people who want to befriend an AI. However, the truth is that we are all being taken advantage of. Taken for a ride. Sometimes, as a way of influencing us to buy a product. Sometimes, just for the creator to be better than you. Smarter. Or is that smarmy?
Can we believe what we see?
AI avatars blur the line between representation and fabrication. When people cannot tell the difference, then this is a deep-fake rather than advertising.
The Advertising Standards Authority (ASA) cracked down on influencers paid to promote company products in posts that were not clearly labelled as advertising. Posts need to include at least #Ad or a similar disclaimer. Yet the ASA’s monitoring in 2025 found that 43% of paid posts (human or virtual) were still missing or burying advertising labels. The risks for companies and agencies are fines, takedowns and loss of reputation.
Meltwater’s social listening found a 50% spike in AI-influencer mentions in early 2025, but sentiment remains polarised. Critics (like me!) question authenticity while brands cheer cost savings.
I assert that the ASA also need to introduce regulations on AI. It is an ethical practice to label an image with a short description to make it accessible to those using screen readers due to a visual impairment. As part of that descriptor, it is good to credit the source of an image. That should be mandatory for AI. For example, “Image of a seagull, AI-generated”. Then it is clear. The descriptor of imagery of the AI Mia Zelu should be similar. Not with “seagull”, of course! The advertising agency Ogilvy has a similar global campaign that requires such influencer posts to be labelled #poweredbyAI. Perhaps the hashtag isn’t dead, but has a new purpose.
Oh. Mia has a ‘sister’ in Ana Zelu. Also a shady AI-generated character. I suspect that the creator, who once mentioned a digital agency, hopes that brands will pay for Mia and Ana to promote products, as has happened with the AI-generated influencers Lil Miquela and Shudu. At the moment, this is at the heart of discussions about AI influencer ethics, accountability and transparency. Fellow PR and communications practitioners, tread carefully!
[AI-generated Image of Mia Zelu at Wimbledon. Source: Instagram @Miazelu]
![[AI-generated Image of Mia Zelu at Wimbledon. Source: Instagram @Miazelu]](https://cornflowercommunications.co.uk/wp-content/uploads/2025/07/screenshot-2025-07-22-at-14.33.24.png?w=1024)



Leave a comment