Part of the AI Regulation News hub.
Brazil's national digital rights secretariat has asked the federal police and the data protection authority to block 80 sites offering AI nudification tools.
The interesting sentence is not the block request. It is the secretariat's position that merely offering a nudification service is inherently abusive, without any showing of a particular victim.
Bottom line: Not binding, and nothing is blocked by it. This is a request from one arm of the executive to the federal police and to ANPD, each of which decides for itself whether and how to act.
Who this affects: Brazilian internet counsel, platform trust and safety teams, criminal lawyers handling image-based abuse, and operators of image generation services reachable from Brazil.
Issue date: Reported by g1 on 18 August 2026. The document's own date is not stated in the report, and no deadline is described.
What changed: The secretariat put a named list of 80 services in front of enforcement bodies and asked for preservation of user access data alongside investigation of operators and users.
Analysis: The request asks for investigation of users, not only operators, and asks that their access data be preserved. That is the part of the story with consequences for people who are not running anything.
Primary sources: g1: Ministerio da Justica pede a PF e a ANPD bloqueio de 80 sites que usam IA para produzir deepnude
- Instrument (EN)
- Request for blocking of 80 nudification services, addressed to the Policia Federal and the ANPD
- Authority
- Secretaria Nacional de Direitos Digitais, Ministerio da Justica; signed by Secretary Victor Oliveira Fernandes
- Jurisdiction
- Brazil
- Status
- Request submitted; no blocking decision reported
- Bindingness
- Non-binding on its face. Any blocking would require action by the addressee bodies.
- Issue date / next deadline
- Reported 18 August 2026; no deadline reported
- Source type
- Trade and general press report quoting the underlying document
- Referral origin
- Complaint by SaferNet Brasil
- Primary source
- https://g1.globo.com/politica/noticia/2026/08/18/ministerio-da-justica-pede-a-pf-e-a-anpd-bloqueio-de-80-sites-que-usam-ia-para-produzir-deepnude.ghtml
What was asked for
According to g1, the Secretaria de Direitos Digitais of the Ministry of Justice asked the Policia Federal and the ANPD to determine the blocking of 80 sites that offer artificial intelligence tools for creating false images and video of nude women.
The document, signed by national digital rights secretary Victor Oliveira Fernandes, also asks for investigation of both the companies offering the services and the users of those services, and for that purpose requests that the users' access data be preserved.
The secretariat states there are indications that real women and even children were victims of the fabrication of nude imagery on these sites.
The legal theory being advanced
The quoted passage that matters reads that the mere availability of services intended for the generation or manipulation of images to create sexualised content of third parties constitutes an intrinsically abusive practice, because it enables the exploitation of the image and identity of people for the production of intimate or sexualised content without their authorisation.
That is a per se argument. It does not depend on identifying a victim, a specific upload, or a failure to respond to a takedown. If enforcement bodies adopt it, the availability of the tool is the violation.
The secretariat pairs it with an access observation: the identified tools operate on the open web, reachable through ordinary browsers, with no registration, identity proof or age verification. That reads as an aggravating factual predicate rather than a separate legal test.
Where the underlying prohibitions come from
g1 reports two separate bases. Altering images of children and adolescents to generate sexualised content is a crime under the Estatuto da Crianca e do Adolescente. Offering tools for creating deepnudes of women is prohibited by a decree signed by President Lula in May.
The report adds that companies operating in Brazil are obliged to block this kind of use of their tools and can be punished if it is shown they did not adopt measures to prevent generation of such content. I have not read that decree, so I am reporting the characterisation, not confirming its terms.
Why the request for user data is the part to watch
Blocking requests against offshore image tools are, in practice, slow and partially effective. The preservation request is different in kind: it aims at Brazilian users of those services, and it is addressed to bodies that can act on it much faster than any blocking order takes hold.
The route the sites were identified matters here too. At least some of the list came from a complaint by SaferNet Brasil, which g1 describes as an NGO working in the defence of human rights on the internet. Civil society referral into a secretariat with the standing to petition the federal police is a workflow worth understanding if you advise platforms.
What we did not verify
I opened and read the g1 report of 18 August 2026 in full, including every passage it quotes from the secretariat's document.
I did not open the document itself, which does not appear to have been published by the Ministry of Justice, and I therefore cannot confirm its date, its number, its addressee formalities or anything it says outside the passages g1 quotes. I did not open the May decree, the ECA provisions, or the SaferNet complaint, and I did not obtain the list of 80 sites.
I will not name any service, state that any site has been blocked, or characterise the legal position of any operator. On the reporting available, no blocking decision has been made by either body.
A ministerial request is a signal about theory, not a change in the law. The theory on offer here is that supplying a nudification capability is itself the abusive act, which would move liability upstream from distribution to availability. Counsel advising image generation services with Brazilian traffic should be tracking whether the PF or ANPD adopt that framing, because the operators are not the only targets named.
Source File
Open the g1 report of 18 August 2026 and confirm that the request came from the Secretaria de Direitos Digitais and is signed by Victor Oliveira Fernandes, that it covers 80 sites and asks for preservation of user access data, and that the report attributes at least part of the list to a SaferNet Brasil complaint.
A mera disponibilizacao de servicos destinados a geracao ou manipulacao de imagens para a criacao de conteudo sexualizado de terceiros constitui pratica intrinsecamente abusiva, por viabilizar a exploracao da imagem e da identidade de pessoas para a producao de conteudos intimos ou sexualizados sem sua autorizacao. ยท Secretaria de Direitos Digitais document, as quoted by g1, 18 August 2026
FAQ
Have the 80 sites been blocked?
Not on the reporting available. The secretariat asked the Policia Federal and the ANPD to determine blocking. Neither body is reported to have decided, and neither is bound by the request.
Are the users of these sites at risk, or only the operators?
The document asks for investigation of both the companies offering the services and the users, and asks that user access data be preserved so that investigation is possible. That is what the report says; whether any user is charged is a separate question the report does not reach.
What makes the conduct unlawful in Brazil already?
g1 identifies two strands: sexualised alteration of images of children and adolescents is a crime under the Estatuto da Crianca e do Adolescente, and offering deepnude tools for adult women is prohibited by a decree the report says President Lula signed in May. The report does not give the decree's number or year. I have not read that decree and am reporting the characterisation.
Is a ministerial request the same as a regulatory order?
No. The Secretaria de Direitos Digitais sits inside the Ministry of Justice and has no power here to block anything itself. The bodies that could act, the federal police and ANPD, retain their own discretion.
Related briefings
Sponsored Training
Practical AI training for regulated professionals, built around verification, documentation and a defensible process. See the courses.