Independent policy desk review · 3 August 2026
AIUndresses.net: AI Undress Policy & Safety Review compares the public rules, privacy disclosures and removal routes of four services. It also explains the current England and Wales position and gives a calm, evidence-preserving response plan for anyone targeted by an intimate deepfake.
01 · Overview
Brand language changes quickly. The durable questions are who operates a service, what it retains, whether consent is explicit and how a subject can obtain removal.
AI undress is a search label for software that alters a clothed photograph so the result appears nude or sexually explicit. The result is not evidence of the subject’s body or conduct. It can still cause real reputational, emotional and economic harm.
An ai undresser or undresser ai may be a website, mobile app or messaging bot. The delivery format does not change the consent problem: a recognisable person has rights even when the output is false.
An undress app can display an 18+ banner while saying little about image retention, operator identity or victim reporting. A policy is evidence of what a provider says, not proof that its controls work as written.
Our review therefore records statements instead of repeating marketing. We did not create accounts, buy credits, upload test photographs or assess output quality. That boundary protects subjects and keeps the comparison reproducible.
The audit can show whether a public document contains a date, an operator, an adult-only rule, a real-person consent rule, a retention statement, a deletion route, a takedown route and processor disclosures.
It cannot certify security, legality in every country or day-to-day enforcement. “Not found” means the item was not located in the documents reviewed on the stated date; it does not prove that no internal measure exists.
People encounter the same category through brand spellings and compressed queries. These terms are included for identification and safety context, not as recommendations:
02 · Original micro-study
AIUndresses.net reviewed eight core documents—one privacy page and one terms page per service—plus the publicly signposted removal routes. The snapshot is dated 3 August 2026.
Method. Each service received one point for each clearly located item: dated policy; registered operator; adult-only rule; explicit consent for images of real people; operational image-retention statement; user-initiated account or full-data deletion; dedicated non-consensual intimate image removal route; and description of third-party processors or disclosures. The denominator is 4 services × 8 checks = 32. Two reviewers using the same pages and rule should be able to reproduce the counts.
Interpretation. This is a transparency audit, not a ranking of image quality, trust or compliance. A “yes” records text on a public page. It does not verify the underlying software, server logs, moderation or age-assurance system. A strict distinction matters because a provider can publish an age rule without deploying an effective age check.
| Service | Policy date |
Registered operator |
18+ rule |
Real-person consent |
Image retention |
Data deletion |
NCII route |
Third parties |
Score |
|---|---|---|---|---|---|---|---|---|---|
| Undress.app | Yes | Yes | Yes | Yes | Yes | — | Yes | Yes | 7/8 |
| Undress.cc | Yes | Yes* | Yes | Yes | Yes | — | Yes | Yes | 7/8 |
| UndressHer AI | Yes | — | Yes | Yes | Yes | Yes | — | Yes | 6/8 |
| UndressAI.io | — | — | Yes | — | — | Yes | — | Yes | 3/8 |
03 · Service document reviews
Names are used for identification. There are no links to generators, pricing calls to action, referral codes or performance claims.
The current terms, last updated 11 January 2026, name UNDR GROUP LIMITED in Belize as operator. They restrict access to adults and prohibit images of minors or real individuals without explicit consent. A separate removal page says a depicted person, guardian or authorised agent can request removal and receive a case ID.
Searches for undress cc lead to a service whose current terms also name UNDR GROUP LIMITED, set an 18+ rule and prohibit use of a person’s image without explicit consent. Its privacy page repeats the 1-hour-to-1-day clearing statement and identifies categories of external providers, including analytics.
The collapsed query undressherai refers to UndressHer AI. Its privacy policy and terms both show an effective date of 15 April 2026. The documents clearly identify uploaded photographs, resulting images and associated metadata as collected media data, and the terms prohibit real-person uploads without explicit consent.
The undressai spelling commonly points to UndressAI.io. Its public privacy page lists personal data, usage data, IP address, browser type, operating system and visited URLs. It states that the products are for people aged 18 or older and that account data can be changed or removed while logged in.
04 · England and Wales
This summary is current to 3 August 2026 and focuses on England and Wales. It is general information, not legal advice.
From 6 February 2026
Section 138 of the Data (Use and Access) Act 2025 brought in offences covering intentional creation, or requesting creation, of a purported intimate image of an adult without consent or a reasonable belief in consent.
As of June 2026
Section 99 of the Crime and Policing Act 2026 targets making, adapting, supplying or offering to supply a thing for use as a purported intimate-image generator. The Act includes defined defences and jurisdiction rules.
Platform duties
Ofcom’s May 2026 statement recommends hash matching for certain services to reduce repeat circulation. The Crime and Policing Act also sets a framework for reports and removal of the same or substantially similar intimate-image content.
The phrase ai undress is broader than any statutory definition. Liability depends on the real conduct, the person depicted, consent, belief, location, intent and applicable exceptions. Images involving anyone under 18 engage separate and especially serious child-protection law. Do not possess, forward or attempt to investigate suspected illegal child sexual abuse material yourself; report it to police or the official child-protection reporting channel for your country.
A useful scale check comes from the Tech Transparency Project. In a 27 January 2026 investigation, it reported finding 55 relevant apps in Google Play and 47 in Apple’s App Store. The identified apps were said to total more than 705 million worldwide downloads and $117 million in revenue, using AppMagic estimates. These are the researcher’s findings, not our independent app-store census, and they underline why store availability should never be mistaken for consent, lawfulness or safety.
05 · Recognise, document, report
Move methodically. The goal is to reduce circulation, preserve evidence and avoid giving an extorter more information or money.
Record where the item appears: the exact page or post URL, account name, platform, date and time. You do not need to prove the depicted body is real. The relevant point is that the content purports to depict an identifiable person intimately.
Take screenshots of the page context, username, messages, threats, report menus and payment demands. Keep originals unchanged and note time zones. Avoid repeatedly downloading or forwarding the intimate image; if a child may be involved, stop and contact the appropriate authorities.
Choose the non-consensual intimate image, impersonation or sexual-content-without-consent category where available. Ask for removal of copies and near-duplicates, preservation of relevant account records, and a case number. Keep confirmation emails and submission receipts.
A host, social network and search engine are different systems. Removal from one does not automatically remove cached results or reposts elsewhere. File distinct reports, quote the original case number and list each URL precisely rather than sending a general homepage address.
Do not pay, negotiate or send more images. In the UK, contact police on 101 for non-emergency advice and 999 if there is immediate danger. Tell the officer it involves an intimate deepfake, non-consensual distribution or sextortion and preserve the demand exactly as received.
Adults can consider a recognised hash-based service such as StopNCII, which creates a digital fingerprint without uploading the intimate image itself to participating platforms. Support from a trusted person, solicitor or specialist helpline can make parallel reporting easier.
Replace brackets with facts. Do not attach an intimate image unless an official, secure reporting process specifically requires it.
06 · Practical prevention
No setting can prevent every misuse of a public photograph, but a smaller, better-controlled footprint gives an attacker fewer high-quality inputs and gives you faster recovery routes.
Limit high-resolution, front-facing photographs on open profiles where that is consistent with your needs. Check old albums, tagged posts and public cloud folders.
Require approval before tagged images appear on your profile. Ask friends and clubs not to publish high-resolution images without checking with you.
Use a reporting email that is not your primary login. This helps you submit notices without exposing the account attackers are trying to access.
Use unique passwords, a password manager and phishing-resistant multi-factor authentication for email, social accounts and cloud photo storage.
Keep backup codes and current recovery contacts offline. A fast account recovery can stop impersonation from becoming a distribution channel.
Record each URL, platform, submission date, case ID and outcome in one document. A clean chronology makes escalation and legal advice more efficient.
If an undressapp or undress ai nude page asks for a test image, leave. Testing with another person’s photo still creates consent, privacy and legal risk.
07 · Device and browser access
This single-page briefing uses a system sans-serif stack, a 16-pixel base size, generous line height and a reading measure of about 68 characters. Headings scale within restrained limits instead of occupying the whole screen. Cards use flexible columns, no fixed heights and aggressive word wrapping, so long brand names and policy terms cannot sit on top of neighbouring text.
The core article is plain semantic HTML and remains usable if JavaScript is disabled. Current Chrome, Edge, Firefox and Safari should receive the full layout on Windows, macOS, Android and iOS; older browsers fall back to a simple vertical document. The comparison table scrolls horizontally on narrow screens rather than shrinking labels into illegibility, and the print control creates a clean paper/PDF version.
08 · Evidence trail
Service claims are attributed to their own public documents. Legal and enforcement claims use official UK sources. External links are marked nofollow and sponsored as requested by the publisher.
28-page decision identifying Undress.cc, the relevant period, age-assurance findings, geo-blocking and penalties.
Official commencement instrument stating that section 138 came into force on 6 February 2026.
Official explanatory material for section 99 on purported intimate-image generators and section 100 removal duties.
18 May 2026 statement on hash matching, later updated 19 June 2026.
27 January 2026 app-store investigation reporting counts, downloads and AppMagic revenue estimates.
4 April 2025 research describing analysis of more than 120 services and crypto-payment patterns.
Service documents reviewed: Undress.app Terms, Privacy Policy and Removal of Non-Consensual Intimate Visual Depictions; Undress.cc Terms, Privacy Policy and public reporting navigation; UndressHer AI Terms of Use and Privacy Policy; UndressAI.io Terms of Use and Privacy Policy. Accessed 3 August 2026. Brand documents may change after publication, so the dated snapshot is essential.
09 · Publisher transparency
Corporate registration establishes who publishes the page. It does not by itself establish legal, safeguarding, forensic or security expertise.
Companies House lists automotive SIC activities: sale of used cars and light motor vehicles (45112), and maintenance and repair of motor vehicles (45200). Those classifications do not demonstrate subject-matter authority in online safety.
The useful E-E-A-T signals here are narrower and verifiable: a named accountable publisher, a dated methodology, direct sources, separation of provider claims from regulator findings, explicit uncertainty and corrections that can be traced to evidence. The article should be judged on that evidence, not on borrowed credentials.
10 · FAQ
Short answers to the questions readers most often need before reporting or interpreting a policy.
No. It is an independent informational review. This page does not accept photographs, create altered images, provide a generator or link readers to one.
No. A written age requirement is a policy statement; highly effective age assurance is a separate technical and compliance question. Ofcom’s Undress.cc decision is a concrete example of why the distinction matters.
Section 138 of the Data (Use and Access) Act 2025 created offences covering intentional creation or requesting creation of a purported intimate image of an adult without consent or reasonable belief in consent; it commenced on 6 February 2026. Other offences and defences may apply, so obtain qualified advice for an individual case.
Keep the exact URL, account handle, date and time, report category, case number, messages and non-explicit screenshots that establish context. Do not repost or circulate the image as proof, and do not download suspected child sexual abuse material.
It means 23 of 32 selected disclosure checks were located in the public documents reviewed on 3 August 2026. It is a documentation score, not a safety, legality, security or service-quality rating.