Brazil Defines Deepfakes Ahead of October Elections
Brazil’s Electoral Court (TSE) is grappling with defining what constitutes a deepfake, just five weeks before the country votes on October 4. The court is expected to rule on broader restrictions for AI-generated campaign material, according to Reuters.
The proposed rules extend beyond simply banning deceptive deepfakes. They include:
- Labeling AI-generated campaign content
- Prohibiting systems that recommend candidates
- Implementing a blackout period during which synthetic political content cannot be published or promoted before and after voting
This definition is crucial because Brazil has encountered a unique case: in July, Jair Bolsonaro (under house arrest) appeared at his son’s campaign launch as an AI avatar identifying itself as a simulation.
While the video was disclosed, it raises legal questions about whether such simulations fall under existing rules targeting fabricated content designed to mislead voters.
The TSE and Supreme Court have expressed differing views on this issue. One justice suggests AI-generated content could be permissible if it doesn’t cause harm, while another warns of potential violations of Bolsonaro’s house arrest and electoral disinformation.
Brazil already had AI regulations in place before this election, introduced ahead of the 2024 municipal elections, which included disclosure requirements for synthetic content.
Tuesday’s ruling is expected to provide clarity, but with only five weeks left until the election, campaigns, platforms, and regulators will have limited time to adapt.
This issue extends beyond Brazil’s borders. Many countries, including the EU, are developing regulations under their AI Acts, assuming that disclosing AI-generated content can reduce its ability to mislead users. However, the Bolsonaro case challenges this assumption because transparent AI simulations can still be politically influential.