AI Act e novas regras que passaram a vigorar a partir de 2 de agosto de 2026

Since the beginning of the month, the AI Act (European Artificial Intelligence Regulation) has entered a new phase, with heightened requirements on issues such as transparency and strengthened enforcement by the AI Office over the provisions of this legislation.

Below are some details on these updates and how they impact Brazilian companies:

• Transparency (Art. 50): users interacting with a chatbot must be clearly informed that they are in contact with an AI, not a human; videos, audio, images generated or manipulated by AI (deepfakes) must be labeled as artificial content; systems for emotion recognition and forms of biometric categorization must inform the affected data subjects about their use.
• Enforcement: the European Commission’s AI Office and national authorities will begin enforcing the provisions of the AI Act and must make formal complaint and reporting mechanisms related to the AI Act available.
• Oversight of general-purpose AI models: obligations for providers of general-purpose models, such as large foundation models, have existed since August 2025, but as of August 2026 the Commission’s enforcement capacity becomes effective in practice.

It is worth noting that obligations for high-risk AI systems did not come into force this August; instead, they will have until 2 December 2027 to comply. For AI systems embedded in regulated products listed in Annex I, the deadline is 2 August 2028.

It is also important to recall that certain AI Act rules have already been in force since the beginning of the year, such as the prohibited practices involving specific manipulative uses and certain forms of social scoring.

How does this impact Brazilian companies?

The AI Act has extraterritorial application, meaning that Brazilian companies are also affected by this legislation when they:

• make AI systems available to users in the European Union;
• use chatbots or assistants targeting the European market;
• distribute synthetic content (text, image, audio, or video) to European consumers;
• offer AI models or services falling within the extraterritorial scope of the AI Act.

In short, organizations must be able to demonstrate where AI is being used, which suppliers support the contracted solutions, and which control mechanisms have been implemented.

What Peck Advogados recommends for companies using AI systems, even if not immediately affected by the AI Act:

• Carry out a corporate inventory of AI systems in use.
• Map critical AI suppliers and their respective contractual responsibilities.
• Implement transparency policies for chatbots and virtual assistants.
• Establish procedures for identifying AI-generated content.
• Integrate AI governance into privacy programs and the Information Security function.
• Monitor the definition of requirements for high-risk systems expected in 2027 and 2028.

For further information, please contact us at contato@peckadv.com.br.

Prepared by: Dr. Graziella Rosa, Head of Digital Advisory, and Dr. Bruno Oliveira, Senior Attorney at Peck Advogados.

AUTHOR

• Bruno Oliveira
• Graziella Rosa