AI-generated images labeled automatically – transparent for everyone.
The EU AI Act demands transparency: whoever uses AI-generated images must be able to disclose it. betterPortrait labels every AI image automatically – from AI selfies to outfit changes – and makes the origin of every photo fully traceable.
Automatic labeling
Every image created or altered by AI automatically carries an AI mark – without anyone having to think about it. AI selfies, outfit changes and all images derived from them are detected reliably.
Complete provenance
The mark travels along through every processing step: if an AI image is cropped, refined or rendered into further formats, the label is preserved. Conventionally edited photos – cropping, background, color correction – remain unmarked.
Visible to everyone involved
Whether in the admin dashboard, in quality control or in the employee portal: AI-generated images are clearly labeled as such everywhere. Your employees always know which version of their portrait is AI – and which is not.
AI only with consent
AI variants are created at the employee’s request – and, if you wish, with a separate consent for AI-edited images. That way you combine openness to innovation with clean governance.
Ready for the EU AI Act
The AI Act introduces transparency obligations for AI-generated content. With betterPortrait you always know which of your employee photos are AI-generated, and you can prove it – the foundation for implementing the requirements cleanly.
Content Credentials (C2PA)
The AI label is written into the delivered files as Content Credentials based on the C2PA standard (ISO/IEC 22144) – cryptographically signed and verifiable by anyone. Downstream systems – from your intranet to your CMS – read the information and display it correctly.
Real photos stay real photos
If you work with a photographer, their images naturally remain unmarked. Only when AI comes into play – an AI variant or an outfit change – is a label applied. The separation stays clean even in mixed libraries.
A clear line to deepfakes
AI-assisted standard editing such as upscaling or relighting changes neither the person nor the meaning of the photo – and does not count as a deepfake under the AI Act. Such images deliberately remain unmarked. That keeps the labeling meaningful instead of stamping everything indiscriminately.
One library, full overview
Original, edited version and AI variant are cleanly linked in the same system. Per person, you see at a glance which versions exist, where they come from and which of them are AI.
Frequently Asked Questions
The AI Act contains transparency obligations for AI-generated and AI-manipulated content. In practice this means companies should know and be able to show which images are AI. betterPortrait provides exactly this foundation – automatic labeling and complete provenance.
The label is created where AI is used and inherited from there through every processing step. There is no manual upkeep – and therefore no forgetting.
No – not even when AI is involved. Our standard editing uses AI in an assistive role, for example for upscaling or relighting. What matters: these steps change neither the person nor the meaning of the photo and count as assistive standard editing under the AI Act, with no labeling obligation. Only what is created or substantially altered by generative AI is marked – for example AI selfies or an outfit change.
No. A deepfake within the meaning of the AI Act is an image that depicts a person deceptively yet falsely. Our standard image editing uses AI in an assistive role – for example for upscaling, relighting or background removal – and changes neither the person nor the meaning of the photo: it still shows the same person, exactly as photographed. The AI Act explicitly exempts such assistive standard editing that does not substantially alter the meaning of the image from the labeling obligation (Art. 50). That is why betterPortrait only labels generative content such as AI selfies or an outfit change.
betterPortrait uses Content Credentials based on the C2PA standard (ISO/IEC 22144) – the open provenance standard backed by Adobe, Microsoft and Google, among others. The label is cryptographically signed and written directly into the image file, verifiable by anyone. This is exactly the mechanism the EU Commission references for machine-readable labeling under the AI Act.
Then the system works for you in the background as a safeguard: all images remain unmarked, and you can prove it at any time. If you later opt for AI features, transparency is guaranteed from the very first image.
Yes. AI variants are an offer: employees choose themselves whether to submit their original or an AI variant. If you wish, the system additionally requires a separate consent for AI-edited images.
Variants that are not chosen are deleted as soon as a variant has been selected. If no image is selected, the selfie and variants are deleted after 7 days, unless you set a different period.
Try it out – no commitment required.
Experience for yourself how easy professional portraits are today.
Try betterPortrait free for 14 days – with three free credits for your first portraits. No commitment, no credit card.