The draft updates the 2023 dosimetry rule for the authority's new powers under the ECA Digital and two decrees on the Marco Civil, and contributions close on Brasil Participativo in 17 days.
Brazil's ANPD ends its first AI sandbox and plans one on children's games
EditorialBy TrustList Editorial
Brazil's data protection authority ANPD closed its first AI regulatory sandbox after testing tools from Prevvine, Metatext and Synapse AI, published a sandbox manual and said the second edition will look at children's digital games.
- Brazil
- Brasilia, Brazil
- Data Privacy Law
- Regulatory Compliance
About Brazil's ANPD ends its first AI sandbox and plans one on children's games
Brazil's ANPD ends its first AI sandbox and plans one on children's games
5 November 2026: Brazil's National Data Protection Authority (ANPD) has finished its pilot regulatory sandbox for artificial intelligence and data protection, publishing the seventh and final monitoring report of the first edition. It also released a Regulatory Sandbox Manual and said the next edition will test how digital games can be made safer for children and adolescents.
Not yet independently verified. Everything here comes from ANPD's own news page published on 5 November 2026 (in Portuguese), read on 19 November. The page gives no opening date or call for the second edition, and the final institutional evaluation report is still being written. What the three tools do is as described on the ANPD page; we did not test them. We will update this when it can be confirmed, and remove this note.
Three AI tools under supervision
The first edition tested three products, with scientific and methodological support from the Centre for Artificial Intelligence and Machine Learning at the University of Sao Paulo:
- STAiDOC, from Prevvine Tecnologia, a clinical decision support tool for doctors. Representatives of Anvisa, the health regulator, said in a preliminary view that a tool of this kind is software as a medical device in a higher risk class, and needs registration or a strict clinical research setting before it is exposed to the public. The test received a favourable opinion with conditions.
- GuardionAI, from Metatext, which protects data flows for companies running autonomous agents and generative AI, tested in simulations for the financial sector. It received a favourable technical recommendation.
- Trajetto, from Synapse AI, which recommends walking routes and was built with the Data and AI Centre at Insper. It received a favourable opinion with conditions and, according to ANPD, scored 86 out of 100 on Unesco's ethical maturity standard.
A manual, not a rule
The new manual explains how ANPD runs a sandbox. The authority stresses that it is guidance, not regulation, and that the rules of the first call, Edital 2/2025, are examples rather than fixed requirements for future editions.
Next: children and digital games
The second edition will be built with the United Nations Development Programme and will follow the Digital Statute of Children and Adolescents (ECA Digital) and its implementing decree. ANPD will first choose an academic or scientific partner; its Innovation Laboratory will then select game developers. The tests will look at ways to curb excessive or compulsive use, including heavy notifications and rewards tied to screen time.
What it means for software companies in Brazil
A sandbox lets a company work through data protection questions with the regulator before a product is in wide use. The first edition showed that the answers can involve more than one authority: the clinical tool raised questions for Anvisa as well as ANPD. Companies building AI for health, finance or other regulated fields in Brazil should expect that kind of joint view.
Game studios and app makers with young users in Brazil should watch for the call for the second edition. Even for those that do not take part, the topics chosen, notification volume and rewards linked to time spent, show where ANPD is likely to look when it applies the ECA Digital rules.
Sources
Categories & features
- Brazil
- Brasilia, Brazil
- Data Privacy Law
- Regulatory Compliance
TrustList Weekly
The week in software and IT, in one email
The news that matters to buyers, new rankings and our own research. Every Thursday, free, and easy to leave.
You may be interested in
ANPD's draft extends its sanctions rulebook beyond the LGPD to the ECA Digital and the new Marco Civil decrees, including age checks. Comments close on 30 November; a public hearing was set for 29 October.
Fines for permanent roaming can start on day 91 of use of a visited network, and clauses that tie mission-critical devices to one network are barred; the list of such uses is due within 60 days.
ANPD's guide for IT product and service suppliers explains the "likely access" test, says a company's own claim about user ages does not decide whether the child protection statute applies, and spells out a proactive duty of prevention.
Anatel technical norm GT-SEG.2026.11 requires mobile operators to verify customers with facial recognition and liveness checks, for new prepaid lines by 30 September 2027 and for all unverified users by the end of 2028.
The Ibero-American Data Protection Network, chaired by Brazil's ANPD, unanimously approved standards on personal data in AI systems, saying public availability of data does not by itself allow its use to train AI.
From 16 November, App Store Connect asks every update whether an app shares real-world location or enables dating; all apps must answer by 12 January 2027. Dating features mean an 18+ rating, and Brazil sets its own ages.
Our news desk records 178 dated obligations from 130 items between 8 October 2026 and 31 March 2027. The week of 26 October is the busiest, with 25 falling due.
More on TrustList
Everything here links back to the same verified catalogue. Pick your next stop.
- More BrazilThe ranking for this subject
- CompaniesAgencies, consultancies and IT service providers, ranked by verified reviews.
- ProductsSoftware and SaaS with pricing, features, integrations and alternatives.
- AwardsAnnual recognition decided by verified reviews and an independent jury.
- LaunchesNew products and releases, voted up by the community every day.
- AI ModelsBenchmark scores and community ratings for every major model.
- RequestsBuyers describe what they need; vendors respond directly.
- PeopleReviewers, authors and makers with public profiles.
- ComparePut up to four listings side by side before you shortlist.