AI Disclosure

Your permission before AI processing

We ask for your explicit permission in the app before sending personal data to the AI providers named in the disclosure. Photo ownership permission, accepting our terms, or reading this policy does not grant AI processing permission. We record the disclosure version and your choice on your account. Declining leaves non-AI features available.

What is sent to AI providers

The photos you select (including visible faces and any crops), your prompts and refinement notes, styling choices such as presentation and region, generated images and image analysis, and an internal account identifier for abuse prevention.

How and why AI data is collected

We collect images you intentionally choose using the photo picker or camera and text and styling preferences you enter or select. We derive image observations and generated results while fulfilling your request. Anthropic, or Anthropic/Google through OpenRouter, analyzes photos and creates styling directions, Style DNA, garment analysis and look sheets. OpenAI checks safety and creates or refines fashion images and model figures.

Third-party protection

We require providers receiving personal data to provide the same or equal protection required by this policy: limited-purpose processing, confidentiality, appropriate security, restricted access and applicable deletion and transfer safeguards. Anthropic handles text and image analysis; OpenRouter routes fallback analysis to Anthropic or Google. OpenAI continues to handle image generation and moderation, and analysis for accounts that have only accepted the earlier OpenAI-only disclosure. Processing is subject to the applicable provider terms and data-processing protections. Provider details: https://privacy.claude.com/ ; https://openrouter.ai/privacy ; https://ai.google.dev/gemini-api/terms ; https://openai.com/policies/data-processing-addendum/

Provider retention and training

We use commercial API services and request providers that do not collect OpenRouter requests for training. Provider abuse-monitoring and legal retention rules still apply; this is not a promise of zero retention. See our privacy policy for provider details. Fashy does not opt your content into model training and disables optional Responses API storage. Retention depends on the provider, feature and applicable safety or legal requirements. See https://developers.openai.com/api/docs/guides/your-data

Withdraw AI permission

Choose Not now to continue without AI. You can withdraw permission in Settings → AI data sharing. This stops new AI requests; work you already requested may finish and data already sent cannot be recalled. To delete source photos, concepts or your account, use the separate deletion controls. Contact [email protected] about data already processed or applicable access and deletion rights.

What the AI does

Fashy sends your selected source photo and styling instructions to an image-generation system to visualize different clothing on you. The system is directed to preserve identity, pose, body proportions, skin tone, and scene while changing clothing and styling, unless you clearly request another permitted edit.

What it does not do

• It does not take body measurements or guarantee exact fit.
• It does not guarantee a garment can be constructed exactly as shown.
• It does not verify fabric, color accuracy, brand, origin, authenticity, price, or availability.
• It is not professional fashion, medical, body-image, safety, or legal advice.

Labels and provenance

Generated images are presented with an “AI concept” label and may include provenance metadata. Do not remove or hide that context in order to mislead. A designer will independently assess feasibility, measurements, materials, price, and production.

Safety and review

Prompts and images may be automatically screened. We block or review suspected non-consensual imagery, sexual content involving minors, nudity, hate, harassment, dangerous or illegal activity, deceptive impersonation, and attempts to evade safeguards. Automated systems can make mistakes; contact support to appeal.