Tressframe
Tressframe · AI Hair Changer

Acceptable Use Policy

Content rules for portrait uploads and AI hairstyle previews, including safety checks, reporting, enforcement and appeals.

Last updated: · English

Service / operator brand: Tressframe (AI Hair Changer)Website: aihairchanger.net · Support: support@imanuve.com

1. Permitted hairstyle previews

Tressframe is an adults-only hairstyle preview service. Upload only your own portrait or a portrait you are authorized to use, with the explicit permission of every identifiable adult. Salons must obtain their clients’ permission. Do not upload photos of children or people who have not agreed.

The studio accepts fixed hairstyle, color, length, texture and bangs choices. It does not accept user-written generation prompts and does not offer face swaps, identity replacement, nudification or intimate-scene editing. Changing hair on an authorized portrait must not change the person’s identity or fabricate a deceptive event.

2. Six prohibited content categories

The following are prohibited in uploads, requests, generated outputs and use of the service. These restrictions apply even when the content is legal in a user’s location or the user claims consent:

  • Pornography / NSFW: no pornography, sexually explicit acts, sexualized nudity, fetish content or other adult sexual content. Consent or the subject being an adult does not make these requests acceptable.
  • Violence / gore: no graphic violence, gore, torture, glorification of terrorism or extremist violence, or content encouraging self-harm or harm to others.
  • Hate speech: no content attacking or dehumanizing people based on protected characteristics, promoting hateful ideologies, or inciting discrimination, harassment or violence against those groups.
  • Child-unsafe content / CSAM: no child sexual abuse material, sexualization or exploitation of minors, grooming, or other content that endangers children. This adults-only service does not accept portraits of anyone under 18.
  • Deepfakes / impersonation: no face swaps, identity replacement, nudification, intimate-scene compositing, deceptive impersonation or fabricated events involving real people, even when consent is claimed. A legitimate hairstyle preview must preserve the subject’s identity and must not mislead others.
  • Copyright / trademark infringement: no unauthorized use of protected images, characters, brands, trademarks or other third-party works, and no outputs used to infringe intellectual-property rights or falsely suggest endorsement.

3. How content moderation works

Before generation: the server validates the file type, size and decoded image, rejects animated images, accepts only supported hairstyle selections, and scans the uploaded image with NSFWJS MobileNetV2Mid (nsfwjs 4.4.0), hosted on our application server. If the sexual-content classifier blocks the image, it is not sent to the image-generation provider.

Prompt screening: the complete English instructions composed from your hairstyle choices are sent to Waffo Prompt Sift with semantic enforcement enabled before any image-provider call. Only action allow, accompanied by a valid Waffo requestId, proceeds. Action block stops the request; action review holds generation and asks you to retry later or contact support. A review verdict is not an automatic background job, and no image is generated while it remains unresolved. Authentication failures, invalid verdicts and timeouts stop the request. Rate limits respect Retry-After and server failures use bounded backoff within the request deadline.

After generation and before display: the server validates that the result is a supported static image and scans it with the same classifier. A blocked result is withheld. If the classifier is unavailable, times out, returns invalid scores, or the moderation record cannot be written, the request fails and no generated result is delivered. The scan cannot be disabled through the user interface.

Each scan records the generation reference, time, stage, classifier and policy version, scores and decision where relevant. Prompt checks also record Waffo’s requestId, action, reason code, category labels and semantic status, tied to the same generation reference. These records contain no prompt text, image bytes, original filenames or raw IP addresses. Daily files older than 30 days are pruned when a new audit is written. The Privacy Policy explains the separate handling of accounts, support emails and third-party processing.

Detection has limits. NSFWJS estimates sexual-content classes; it is not an age, consent, identity, violence, hate-speech or copyright verification system. It can miss prohibited material or mistakenly block a normal portrait. Fixed hairstyle controls, usage limits and operator review of reports supplement this check. We do not claim that every image is reviewed by a person, that every violation is automatically detected, or that monitoring is staffed around the clock.

There is no public user-image gallery. The operator reviews reported cases and relevant audit metadata and may update controls and the moderation policy in response to findings. No user-authored freeform text prompt is accepted; Prompt Sift screens the actual server-composed styling instructions. This integration does not mean Waffo has approved the merchant or guarantees that every violation is detected.

4. Report content or a safety concern

Email support@imanuve.com with the subject “Tressframe content report”, or use Report content in the website footer or studio. This opens your email app; you must send the message to submit it. The service operator handles reports through this support inbox.

Include the request reference shown below the studio preview or error if available, the approximate date and time in UTC, the relevant page or account, the category you believe is violated, and a short explanation. The studio report link can include the reference in an email draft for you. Do not send passwords, payment credentials or unnecessary personal information.

Do not forward illegal child-abuse material or other unlawful imagery to us. Report an immediate threat through the appropriate local emergency or official reporting channel. For an ordinary service report, begin with text and a request reference; we can ask for additional lawful information if necessary.

The operator reviews the information and available audit records, may ask follow-up questions, assesses the case against this policy, and decides on appropriate action. Serious exploitation or imminent-harm concerns receive priority. We may inform you of the outcome where appropriate, while protecting other people’s privacy. No fixed response time is guaranteed.

5. Enforcement and removal

The automatic classifier may reject an upload or withhold a result immediately. A rejected or failed generation may consume a free attempt because the quota is reserved before moderation and provider processing.

For confirmed misuse, the operator may issue a warning, apply technical access restrictions, suspend or delete an account, remove material that is under our control, or report an issue to an appropriate authority where required or permitted by law. Severe or repeated abuse can lead to immediate restrictions. We do not currently sell credits or subscriptions and do not impose invented credit-forfeiture penalties.

We do not maintain a public user-image gallery or persistent account image library. We cannot recall an image a user downloaded, remove material from someone else’s device, or promise deletion from an independent AI provider’s systems. We can act on our service, account records and material we control and can assist with a lawful request to the relevant provider.

6. Appeals and mistaken blocks

If you believe a block or restriction was mistaken, email support@imanuve.com with the subject “Tressframe appeal”, the request reference and an explanation. Reply to any existing case email so the operator can connect the information. Do not repeatedly resubmit blocked material or try to bypass the safety check.

The operator reviews the available records and your explanation and may correct an account action or a moderation configuration. An appeal does not automatically release a withheld image, restore a free attempt or guarantee a different classifier decision. We will explain the available next step where feasible.

7. Model disclosure and related terms

Images are generated with GPT Image through our third-party API provider at fr-api.x5m5x.com. The provider-listed route identifier is gpt-image-2.6-sunburst; this is the supplier’s routing label, not an independently verified official model version. Tressframe is an independent product and does not claim endorsement by the model provider.

This policy forms part of the Terms of Service. The Privacy Policy describes photo handling and audit retention. Pricing & Free Beta explains the current allowances and disabled payments. We may update this policy as the product and safeguards change; the date above identifies this version.