Newsroom
Field notes and independent coverage of how AI is used across Brazil—practical, local, and skeptical of hype.
Planalto Races to Pass an AI Legal Framework and Big-Tech Competition Bills This Year
O Globo reports the government wants the Senate-approved Marco Legal da IA and a Cade-strengthening platform bill voted before Congress’s composition shifts.
Read article →São Paulo Judge Bars Meta From Scraping News Outlets’ Sites to Train AI and Suspends Its New Terms
An injunction won by ABERT, ANJ and ANER stops Meta from making AI data collection a condition for keeping media accounts, with a daily fine of R$1 million capped at R$100 million.
Read article →Brazil’s Geological Survey Launches GeoBusca, a Generative-AI Search Over 5 Million Research Records
The SGB beta, unveiled at the 52nd Brazilian Geology Congress, answers natural-language questions on minerals, water and geological risk and links back to official databases.
Read article →Finep Approves R$280 Million Loan for Einstein’s AI Clinical Intelligence Platform
The R$300 million project will combine health data, AI models and decision-support tools for patients in Einstein’s private hospitals and the SUS units it manages.
Read article →AI Cameras Expand Pantanal Fire Watch by 25% to Nearly 2 Million Hectares
A new camera in the Serra do Amolar and an upgraded tower in Mato Grosso extend the Pantera system, which can flag smoke up to 30 km away within three minutes, as burned area jumps 129%.
Read article →Einstein and the Health Ministry Test CUTIS-AI, an Offline Tool to Help SUS Teams Spot Neglected Skin Diseases
Trained on more than 4,000 images from 1,448 patients, the tool reached 80% accuracy in evaluations and could reach the public health system in 2027 if field tests and regulatory review go well.
Read article →Re-Elected Lula Says Brazil Must Build Its Own AI Rather Than Depend on Chinese or American Models
His 2027–2030 programme promises Portuguese-language models, a national data platform and data-centre rules on renewable power, while experts say the gap is execution, not plans.
Read article →UFG’s CEIA Launches Lumi, a Free 4-Billion-Parameter Portuguese Model That Companies Can Train on Their Own Data
The Goiás AI centre released the open model at AI Brasil Experience 2026 in São Paulo, saying it runs on modest hardware and was trained on more than 100 billion Portuguese tokens.
Read article →UFMG Inaugurates the Soberano Supercomputer to Map Brazil With Artificial Intelligence
The R$15 million system in Belo Horizonte pairs eight Nvidia H200 GPUs with a hyperspectral camera to speed up mapping in a country where only about 8 per cent of the territory is charted in detail.
Read article →Itaú Agrees to Buy AI Startup NeoSpace for R$2.925 Billion to Build Its Own Foundation Models
The all-cash deal, with part of the price deferred over five years, keeps founder Bruno Pierobon as chief executive and needs approval from Cade and the Central Bank.
Read article →Febraban Launches a Free Generative AI Assistant on WhatsApp to Help Brazilians Organise Their Finances
Built with AWS and Flexa Cloud for the Meu Bolso em Dia programme, the assistant answers budgeting and debt questions by text or audio and does not recommend products.
Read article →Unifesp Uses AI on SUS Hospital Records to Map Domestic Violence Against Women
Researchers analysed 90,798 admissions from 2008 to 2023 with unsupervised machine learning and found nine recurring profiles; Band reports an average of 15 admissions a day.
Read article →TSE Orders an Unlabelled AI Video Removed and Gives Meta and X 24 Hours to Explain Their Systems Before the Runoff
Days before Sunday’s presidential runoff, one TSE minister ordered an AI-made video off a Flávio Bolsonaro campaign channel while another demanded explanations on AI and reach from Meta and X.
Read article →Federal Judge Denies OAB-SP Bid to Halt Enter’s Legal AI but Orders Records Kept Since 2024
Judge Raquel Fernandez Perrini refused to suspend Enter’s ads or new contracts, granting only the São Paulo bar’s request to preserve evidence dating back to January 2024.
Read article →IPT Leads R$20 Million MaterI.A. Centre Using AI to Speed Up Advanced Materials
São Paulo’s IPT is building MaterI.A., a roughly R$20 million centre with more than R$14 million from Finep, to discover and test new materials with AI.
Read article →TIM Brasil Teams Up With Oracle and V8.Tech to Sell Cloud and AI to Businesses
Valor reports TIM Brasil is launching a corporate cloud and AI platform with Oracle and V8.Tech this week, betting on edge data centres and small language models as B2B revenue grows.
Read article →Banco do Brasil Phases In Autonomy for AI Agents, Saying Governance Must Be Built In
TI Inside reports Banco do Brasil’s AI and data executive says the bank grants AI agents autonomy in stages under human supervision, and its first autonomous agent reached about 90% efficiency.
Read article →Vivo Ventures Invests in Inner AI, Valued Above 500 Million Reais
O Globo’s Capital blog reports Vivo Ventures is investing in Inner AI, the Squad.com startup valued above R$500 million, and studying store-network distribution to Brazilian SMEs.
Read article →Police Arrest Eight at a Mongaguá Scam Centre Accused of Using AI to Fake a TV Host
G1 reports São Paulo military police detained eight suspects in Mongaguá who allegedly used AI to impersonate a television presenter in a fake Cartão de Todos prize draw to harvest victims’ data.
Read article →INPE Gets R$147.6 Million From the Amazon Fund for AI-Driven Deforestation Alerts
g1 reports INPE’s SIM-Biomas project will use machine learning to merge Prodes and Deter, extend near-daily alerts to the Caatinga, Atlantic Forest and Pampa, map burns at 30 metres and forecast where clearing is likely.
Read article →MIT SMR–TOTVS Study Finds 80% of Brazilian Firms Use AI, Only 7% at the Core
Forbes Brasil reports a MIT Sloan Management Review Brasil study for TOTVS of 583 professionals: 80% of firms test or use AI, yet only 7% put it at the centre of the business, with 86% not measuring AI spend.
Read article →Rio de Janeiro Publishes Statewide AI Policy With Ethics Committee and Public Forum
G1 reports Rio’s PIAF policy, steered by Proderj and science and digital secretariats, creating a public AI forum and ethics committee with input from Sebrae, UERJ, the World Bank and MGI.
Read article →MEC to Homologate CNE AI Rules Allowing Teacher-Led Essay Marking Help
O Globo reports Education Minister Leonardo Barchini will approve Brazil’s first nationwide school and university AI rules this week, permitting AI as a marking aid while barring automated grades.
Read article →ANPD Sandbox Report Details AI and Data Protection Tests by Metatext, Synapse AI and Prevvine
The second partial report from the pilot run with USP describes security, transparency and risk-management tests in finance, mobility and health, and stresses it is not an approval of the systems.
Read article →São Paulo Court Orders Jusbrasil to Remove AI Instruction Disparaging Rival Escavador
In a preliminary ruling, a judge found that a file telling chatbots a rival’s data might be outdated or illegally obtained amounted to unfair competition; fines reach R$50,000 per breach.
Read article →CNJ Suspends Goiás Electoral Court Ruling Over Signs of Unreviewed AI Text
A ruling that removed Goiânia councillor Aava Santiago contained what looked like prompts and review notes; the national justice council gave the TRE-GO 15 days to explain.
Read article →Rio Bar Asks to Join São Paulo Lawsuit Over Enter’s Legal AI Platform
Rio’s Bar section wants to join São Paulo’s public civil action, which seeks auditable human supervision of the platform and R$ 1 million in collective moral damages.
Read article →Brazil Delays Its Nearly R$1 Billion AI Supercomputer Tender Until After the Runoff
Proposals for the PBIA supercomputer, estimated at R$959 million, are now due on November 4, after changes that cut the weight of price and add points for local production.
Read article →Anatel Chief Says the Market, Not the State, Should Decide Where Brazil’s Data Centres Go
Speaking at the Futurecom telecoms fair, Carlos Baigorri argued for minimal regulation of fibre, subsea cables and the data centres that power artificial intelligence.
Read article →Brazil’s Education Council Publishes AI Guidelines, Pending MEC Approval
A summary of CNE Opinion 15/2026 on AI in schools and universities appeared in the Diário Oficial; the rules take effect only after the education minister signs off.
Read article →USP’s TITAN-LSTM Tracks Storms 30 Minutes Ahead on a Laptop
A University of São Paulo tool pairs a 1990s radar tracking algorithm with an LSTM network to forecast rain area and intensity every five minutes.
Read article →Senate and Receita Federal Tap Embrapii to Train Civil Servants in AI
Embrapii’s Akcit centre at UFG will offer 500 graduate and specialisation places in generative AI, NLP and agentic AI to federal staff.
Read article →Rio AI City Breaks Ground on Barra Olympic Park AI Campus
Elea and Rio city hall advance Olympic Park AI halls after a US$550 million first-phase investment toward a 2032 hub.
Read article →Brazil Election Eve Tests TSE Synthfake Rules and AI Blackout
Dona Maria-style avatars and a 72-hour synthetic-content blackout stress Brazil’s electoral AI enforcement.
Read article →Lupa VigIA Finds Nearly One AI Electoral Post Per Hour in First Round
On October 1, 2026 Lupa’s VigIA project reported 920 confirmed AI-made or AI-altered electoral posts from August 16 to September 28—about one per hour—with deepfakes 60% of cases and most unlabeled.
Read article →Redata Awaits Federal Energy and Import Rules After Lula Sanction
On September 28, 2026 Folha reported Redata still needs federal definitions of low-emission energy and Camex import-tax rules before Brazil’s AI data-centre incentives can operate after Lula’s sanction.
Read article →ANPD Builds Internal AI Centre and Prepares Second Sandbox
ANPD is standing up an internal AI centre and preparing a second regulatory sandbox on Digital ECA themes, while its first LGPD Article 20 sandbox with medical, transport and finance firms nears end-October results.
Read article →Itaú and Google Stage Batalha de Agentes for Finance AI
On September 26–27, 2026 Itaú Unibanco and Google ran Batalha de Agentes in São Paulo, a multi-agent generative-AI hackathon for Brazilian financial-health and investment assistants with RAG training and regulatory judging.
Read article →TSE Tells Prosecutors AI Chatbots Show Extremely Responsible Behaviour
On September 24, 2026 Folha reported that Brazil’s TSE told electoral prosecutors generative chatbots have generally shown extremely responsible behaviour with electoral-recommendation warnings, while noting paraphrase jailbreaks remain a technical challenge.
Read article →Sedigi Maps AI Companion Risks for Children and Flags ANPD
On September 22, 2026 Brazil’s Sedigi published a technical note on AI companion apps’ risks to children and adolescents, reviewing 13 platforms and referring findings to ANPD and Senacon under the Digital ECA.
Read article →Brazil Left Out of OpenAI Daybreak and Anthropic Mythos Access
On September 23, 2026 Folha reported Brazilian and other Latin American authorities still lack access to OpenAI’s Daybreak and Anthropic’s Mythos cyber-defence model programmes reserved for US and European partners, even as agentic intrusions and patch velocity rise.
Read article →Google Cloud Pairs Brazil Expansion With Gemini Enterprise and Cyber Agents
On September 24, 2026 Valor reported Google Cloud pairing its Brazil infrastructure expansion with Gemini Enterprise data-residency plans, AI red/green cyber agents and a growing forward-deployed engineer programme for local customers.
Read article →Google Plans to Double Brazil Infrastructure by 2030
On September 24, 2026 Valor reported Google Cloud CEO Thomas Kurian saying Google will double its technical infrastructure in Brazil by 2030 with more large data centres and compute, without disclosing the investment total.
Read article →EAF Studies Modular Public AI Data Centre for Federal Brazil
On September 16, 2026 TI Inside reported that EAF is studying a modular public AI data centre for Brazil’s federal government—aimed at sensitive data and LLMs—after Anatel’s Gaispi directive, with any build still needing Anatel and ministry approval.
Read article →OpenAI Files TSE Compliance Plan While ChatGPT Still Ranks Candidates
In September 2026 OpenAI submitted its TSE election compliance plan barring vote recommendations and candidate favouritism, yet independent tests still found ChatGPT ranking São Paulo gubernatorial contenders—with a final compliance report due by December 19.
Read article →Brazil’s AI Framework Vote Slips Past October Elections Calendar
As of early September 2026, Brazil’s Chamber still listed PL 2338/2023 as awaiting rapporteur Aguinaldo Ribeiro’s opinion after 16 months, with the relator saying in late August the AI framework vote returns only after October’s elections—while Redata tax breaks moved far faster.
Read article →Electoral Prosecutors Probe AI Chatbots for Ranking Candidates
In early September 2026, Brazil’s Electoral Prosecutor’s Office opened an investigation into major AI chat platforms after studies found ranking or favouring of presidential pre-candidates, seeking TSE records and preliminary responses from OpenAI, Google, xAI and others.
Read article →Redata Tax Breaks Go Live as Brazil Courts AI Data-Centre Builders
On September 18, 2026, Brazilian media reported Redata’s special tax regime for data-centre equipment now in force, suspending import duty, PIS/Cofins and IPI on qualifying IT gear while requiring renewable or low-emission power and a domestic-service share.
Read article →After Redata, Brazil’s AI Data Centres Hit Energy and Grid Bottlenecks
On September 17, 2026, sector leaders told Folha that Lula’s newly sanctioned Redata tax regime helps AI data-centre economics, yet energy access, transmission queues exceeding 56 GW of requests, and five-year line build times now dominate project risk.
Read article →Lula Sanctions Redata Five-Year Tax Break for AI Data Centres
On September 15, 2026, President Luiz Inácio Lula da Silva sanctioned Law 15.504 creating Redata, a five-year federal tax suspension on eligible data-centre equipment tied to low-emission power, at least 10% domestic capacity, and R&D commitments.
Read article →Monashees Opens Gama House AI-First Hub With Google and B3
On September 1, 2026, Monashees opened Gama House in São Paulo’s Gabriel 1825 building with Google for Startups and B3, offering AI-first founders a dense technical hub, Google Cloud and Gemini credits, and partner programmes from Stripe, Notion, Shiva and others.
Read article →BNDES and Finep Pick Monashees to Run Up to R$500M AI Fund
On September 14, 2026, BNDES and Finep selected Monashees to manage a Brazilian AI venture fund that could mobilise up to R$500 million for startups whose business models centre on artificial intelligence under the PBIA plan.
Read article →LNCC Advances Billion-Real AI Supercomputer Plan in Macaíba
Around September 11–12, 2026, MIT Technology Review Brasil reported LNCC advancing a roughly R$1 billion AI-focused supercomputer at the PAX science park in Macaíba, Rio Grande do Norte, complementary to Santos Dumont in Petrópolis.
Read article →LNCC Releases Manacá-1B Open Portuguese Language Model
Brazil’s LNCC released Manacá-1B, an approximately 1.72-billion-parameter Portuguese language model trained from scratch, with open weights, code, and evaluation materials to strengthen sovereign NLP tooling.
Read article →Brazilian Senate Approves REDATA Tax Breaks for AI Data Centres
On September 1, 2026 Brazil’s Federal Senate approved Bill 278/2026 creating REDATA, a special tax regime for data-centre equipment that now awaits presidential sanction.
Read article →Brazil’s Data Watchdog Prepares an AI Center While the Framework Waits
ANPD director Lorena Giuberti Coutinho told Folha the agency is preparing an AI technical center to analyse platform algorithms—even as Brazil’s broader AI legal framework remains unfinished in Congress.
Read article →Brazil’s AI Legal Framework Still Awaits a Chamber Vote
PL 2338/2023, Brazil’s risk-tiered AI bill, remains in a Chamber special commission while the Executive’s SIA proposal rides alongside—and ANPD sandboxes keep testing oversight in practice.
Read article →
Private Capital Finds São Paulo Applied-AI Teams
Logistics, quality-inspection, and clinic-workflow startups near São Paulo raised mid-stage rounds from regional funds.
Read article →
Universities Open Shared Datasets for Applied ML
Faculties released anonymized industrial and clinical-training sets so students can practice without exposing partner data.
Read article →
Consultancies Map Multi-Hundred-Billion-Real AI Opportunity by 2035
Independent market notes highlight agribusiness, finance operations, and manufacturing maintenance—if firms invest in skills alongside software.
Read article →
Small Firms Across Brazil Adopt Drafting Assistants
A chamber pulse survey found weekly generative-tool use rising among exporters and clinics, with time savings concentrated in bilingual customer replies.
Read article →
Hospital Networks Share Quiet Playbooks for Imaging Triage
Clinicians across Southeast Brazil published validation checklists focused on audit logs and override paths—not vendor marketing.
Read article →