How we review and how the scores are built

Privacy carries the largest weight because it is the criterion a reader cannot verify for themselves after the fact. By the time retention matters, the file has already been uploaded.

Two sources are admissible. The first is what the provider publishes on its own public pages: prices, limits, retention periods, company details. The second is what the interface does in front of us in the standardised scenarios. Nothing else enters a score.

Marketing copy is not evidence of a capability. A claim that images are deleted immediately, with no period stated and no mechanism described, is recorded as a claim and not as a retention policy. A provider that publishes nothing on a criterion receives the mark that an absence earns, which is low, rather than a guess.

Every figure carries the date it was verified. Prices in this category change frequently and without notice, and a number without a date is not information. When the desk returns to a tool, the new date replaces the old one and the score moves if the facts moved.

The same test set is used for every tool: well-lit frontal photographs at medium and high resolution, taken from our own material. Each tool is run on the free tier where one exists and on the lowest paid tier, and the output is examined at full size rather than at the size the interface displays.

What we never do: we do not test these tools on photographs of third parties, we do not publish results from such attempts if they reach us, and we do not accept material from readers for testing. A review desk that built its evidence base that way would be committing the offence it writes about.

Six of the fourteen entries carry an affiliate link and eight do not. The links are marked on every page that carries one, they earn a commission at no cost to the reader, and they have no effect on a score. Two of the lowest-scoring entries in the catalogue carry affiliate links, and the highest-scoring entry would hold its position without one.

No provider reviews text before publication, no provider is offered a right of reply that changes a mark, and no placement in the ranking is for sale. Where a factual error is pointed out, it is corrected and the date on the affected figure changes.

Lawful use only with your own photos, or photos of adults who have given explicit consent. Responsible AI (in Spanish).

• Rombach et al., High-Resolution Image Synthesis with Latent Diffusion Models — The paper behind the diffusion inpainting that every tool in this catalogue relies on.
• Kirillov et al., Segment Anything — The segmentation approach used to find a garment region automatically.
• Diario Oficial de la Federación, Ley Olimpia reforms — The Mexican text criminalising dissemination of intimate content, real or simulated, without consent.
• Regulation (EU) 2024/1689, the AI Act — Article 50 sets the disclosure duty for synthetic content that this site follows.
• StopNCII.org — The international route for having non-consensual intimate images removed.
• TAKE IT DOWN Act, S.146 — The United States takedown obligation, relevant where a provider is hosted there.

Technical claims on this page rest on the published research above; legal claims rest on the published texts. Everything else is what the desk observed in its own standardised scenarios, with the date recorded beside each figure.