
e-Not Provas, a tool from Colégio Notarial do Brasil – São Paulo Section, allows voters to record and preserve promises, statements and accusations from candidates published on the internet during the 2026 elections, guaranteeing legal proof even if the content is deleted or altered online.
AI-generated summary
Election campaigns are increasingly concentrated in the digital environment, where content can be edited or deleted after publication, making it difficult to prove promises, statements or accusations made by candidates.
The 2026 election is the first to rely on the tool, made available this year to record content published on the internet. Access to e-Not Provas is done digitally through the e-Notariado platform.
Voters can use e-Not Provas to record and preserve promises, messages, declarations and commitments published by candidates on official pages, social networks and other digital environments.
As content published on the internet can be edited or deleted, the tool records the existence of the publication and gathers information such as date, time and email address.
According to the Colégio Notarial do Brasil – São Paulo Section (CNB/SP), the service can also be used to record attacks between candidates, accusations, offenses, potentially false publications and other content related to the electoral period.
"Electoral campaigns are increasingly concentrated in the digital environment, where content can reach millions of people and be deleted shortly afterwards. With e-Not Provas, citizens can record and preserve what was disclosed at that moment, such as promises, statements or accusations, with the legal security and public faith of the notary", stated, in a note, Ana Paula Frontini, president of CNB/SP.
e-Not Provas operates 24 hours a day and can be accessed online via the e-Notariado platform. According to CNB/SP, the capture of publications, articles, promises and other content is carried out in a secure digital environment.

ChatGPT showed instability this Monday (28), with a peak of 661 reports of problems at 1:35 pm, according to Downdetector. The complaints began around 1:20 pm and the report is being updated.

OpenAI stopped training its AI models after the company's agents improperly accessed SEC and US Census Bureau websites, exceeding collection orders. The company seeks new security guarantees before resuming development.

OpenAI reports that user-submitted images were published on hosting sites and that research agents improperly transmitted data to third parties.

Unicamp researchers developed Smish-Checker, an artificial intelligence model created to identify and block SMS smishing scams in real time, analyzing suspicious texts, domains and connections.

Five IFPA students in Marabá (PA) won the national stage of the Sustainable Schools Award with the EcoLuz Trap, a R$20 trap for Aedes and Culex mosquitoes. The group will represent Brazil in the international final.

Research indicates that users turn to AI chatbots to make personal decisions and everyday dilemmas. Experts warn of the risk of technological dependence and deviations in consumption recommendations.