Zara, H&M and Zalando are now using artificial intelligence to dress models virtually from a single photo shoot. Behind the brands’ efficiency pitch, professionals are speaking out about non-consensual use of their likeness and AI-generated face replacements carried out without their agreement.
Key Takeaways
- Zara generates hundreds of product images using AI from a single photo session per model
- Models testify to unauthorized use of their likeness and AI-driven face replacements without their consent
- The EU AI Act imposes no general transparency obligation on the use of AI in advertising campaigns
Have an AI Sum Up This Article
ChatGPTZara, H&M, Zalando: The Race for Low-Cost Imagery
Zara’s approach follows a straightforward industrial logic. A model poses during a single photo session. AI then dresses them virtually in hundreds of different items, with no additional physical presence required. According to an anonymous account, models are compensated at a rate equivalent to what they would earn for additional photo sessions.
An Inditex spokesperson, Zara’s parent company, stated the group uses AI “only to complement existing processes.” H&M announced it is using AI model cloning for marketing purposes. Zalando is also deploying AI to create catalog images faster. All three frame the technology as a tool that supports creative teams rather than replacing them.
The institutional messaging is consistent across all brands. AI is positioned as a complement, not a substitute. The on-the-ground experience described by working professionals tells a different story.
The economic logic is direct. Fewer photo sessions, fewer travel costs, fewer crews to mobilize. For large retailers producing thousands of visuals per season, the savings are substantial. The pressure on production budgets has intensified with the rise of e-commerce, and AI represents a technical answer to the volume challenge.
As we covered in our analysis of ClickUp’s mass layoffs, the complementarity argument between AI and human teams is consistently deployed by companies reducing their production costs. The fashion industry is following the same pattern.
Faces Replaced, Rights Ignored
French model Audrey Soligny returned from a shoot in India to discover that her images had been used without her consent to generate AI visuals. Checking a brand’s Instagram account, she found photos she had never taken, built from the plain white-background shots captured that day. Her visual data had been exploited without her knowledge by an agency operating out of India, a country where, she noted, “there is absolutely no concept of image rights,” even when the resulting visuals circulate internationally.
Charlotte Lemay’s account describes an even more radical practice. On arriving at a shoot, she was told there would be no makeup or hairstyling, because AI would generate a new face on her silhouette. The model’s image was no longer being used. Her body was repurposed as an interchangeable canvas, stripped of its visual identity.
The economic dimension is explicit in the testimonies. Audrey Soligny points to the concrete consequences: a model whose face is replaced by AI gradually stops being called back for new shoots. Photographers, hairstylists and makeup artists lose work too. AI does not eliminate a single position. It cuts volume across an entire chain of professionals.
The British Association of Photographers has formalized this warning, stating that AI will structurally reduce the number of available assignments for photographers, models and production crews in the years ahead. This is no longer a theoretical projection. The practices described by Audrey Soligny and Charlotte Lemay are concrete, documented illustrations of it.
Also on Horizon:
- DuckDuckGo Up 30%: Users Reject Google’s Forced AI Search
- Microsoft Drops Claude Code After Burning Its Entire AI Budget
- Meta Blocks Apple Intelligence and Challenges Apple on AI
A Legal Framework That Cannot Keep Up
The EU AI Act was expected to introduce safeguards. Its implementation is phased, and its scope in the advertising sector remains limited. Katrien Maris, a media and communications lawyer, confirms there is no general obligation to disclose AI use in advertising campaigns. Exceptions apply to deepfakes, AI impersonations of identifiable real people, and chatbots.
Karine Ysebrant, AI expert at BAM (the Belgian Association of Marketing), describes AI use that now spans “every level of message creation: from market data processing to the creative stage, internal projects, and the final output, in voice or image.” Making labeling mandatory would mean labeling constantly, without specifying what type of AI is used or for what purpose.
The International Chamber of Commerce (ICC), cited by Karine Ysebrant, takes a more targeted position: AI should be disclosed “when it has transformative power for the public, when it can be harmful if not mentioned.” A formulation that leaves advertisers considerable room for interpretation.
That tolerance has already been tested in the Netherlands. McDonald’s was forced to pull its Christmas television ad after the video, produced entirely with AI, was widely criticized by consumers as artificial and excessively clichéd. A public pushback that shows audiences are not yet ready to accept AI without transparency, even in the absence of a legal requirement to disclose it.
In the short term, brands will continue to operate in the current legal gray area. Over the medium term, pressure from affected professionals and informed consumers could push legislators to clarify consent and transparency obligations around the use of image data in AI. The AI Act has the conceptual tools to do it. The political will to apply them to the advertising sector has yet to materialize.
Follow the story on Horizon.
