Brazil already has some of the world’s more explicit rules governing AI in elections. An AI-generated version of former President Jair Bolsonaro is now testing how far those rules actually reach.
The Superior Electoral Court is confronting questions over a video that recreated Bolsonaro’s face and voice to support the presidential campaign of his son, Flávio Bolsonaro. Because the former president is under house arrest and subject to restrictions on public communication, the case raises a difficult question: can a synthetic likeness effectively speak where the real person cannot?
The dispute arrives as AI-generated political content spreads online and voters increasingly turn to chatbots for information about candidates. Brazil may have written rules for election AI, but enforcing them is becoming a much harder problem.
AI is testing an electoral gray line that only the courts can draw #
The problem for Brazil’s election authorities is not that AI was used in a campaign. The problem is how subtly AI can cross established lines without appearing to have done so. And the use of an AI-generated Jair Bolsonaro has shown a clear example of that.
In July, Flávio Bolsonaro’s campaign used an AI-generated avatar of his father, former President Jair Bolsonaro, during a party convention.
Because the real Bolsonaro is under house arrest and subject to judicial restrictions prohibiting public statements directly or via third parties, the case raised a question that could not be settled by simply asking whether the video was labeled as AI.
While the TSE cleared Flávio Bolsonaro’s campaign on electoral grounds, Supreme Court justices continue to evaluate whether third-party deployment of an avatar violates the conditions of Jair Bolsonaro’s sentence. Did the synthetic version effectively allow his political message to reach voters through a machine-generated likeness, even without his direct authorization? That is the thin legal line the Superior Electoral Court now has to interpret in the weeks before the election commences.
Brazil wants AI, but with electoral limits #
Under the TSE’s rules, AI isn’t going away entirely, but its presence must be subject to tighter limits.
The rules prohibit deepfakes, including synthetic content used to create or alter a person’s image or voice to spread false information or harm a candidacy. AI systems are also barred from ranking or recommending candidates, parties or campaigns, even when a voter explicitly asks them to do so.
The court has also created a tighter window around election day. AI-generated or altered content cannot be used during the 72 hours before voting through the 24 hours after the election ends.
To enforce these rules, Brazil is also requesting that providers adhere to the country’s AI laws and file compliance plans during the election period.
The result is a system that treats the use of AI itself as permissible but not unlimited.
What's hot at TechRepublic
[Blackpoint Cyber vs. Arctic Wolf: Which MDR Solution is Right for You?](https://www.techrepublic.com/article/blackpoint-cyber-vs-arctic-wolf-which-mdr-solution-is-right-for-you/) -
[Why AWS Sellers Choose Deepgram Over Other Voice AI Tools](https://www.techrepublic.com/article/why-aws-sellers-choose-deepgram-over-other-voice-ai-tools/) -
[SS&C Intralinks DealCentre AI vs. Datasite: Which platform is built for the future of dealmaking?](https://www.techrepublic.com/article/dealcentre-ai-vs-datasite/) -
[SS&C Intralinks FundCentre AI vs. Juniper Square: Which platform better supports modern private markets fund managers?](https://www.techrepublic.com/article/fundcentre-ai-vs-juniper-square/) -
[Verito vs. Rightworks: Which IT Provider Is Best for Your Firm?](https://www.techrepublic.com/article/verito-vs-rightworks-which-it-provider-is-best-for-your-firm/)
Possible ways AI can interfere with elections #
People increasingly use AI chatbots as shortcuts for information, but attackers can manipulate the material those systems retrieve and summarize.
A February demonstration by Thomas Germain showed how AI tools can surface a fake blog post, raising the possibility that political actors could use manipulated web content to influence what voters are told.
Brazil’s rules prohibit AI systems from ranking or recommending candidates, but popular tools are already doing it.
Reuters report, citing ITS Rio, reveals that seven major AI tools in Brazil ranked political opponents in 90% of the responses it tested, meaning a voter looking for information could instead receive an AI-generated judgment about who deserves their vote.
Generative AI can dramatically increase the volume of political misinformation an actor can produce.
A single campaign could generate thousands of slightly different messages, videos, images, or narratives for different audiences, making coordinated misinformation harder for platforms and fact-checkers to identify as one operation.
What Brazil’s election situation could mean for everyone #
Brazil offers an early case study in a problem other governments are likely to face: writing AI election rules is easier than enforcing them once synthetic candidates, manipulated information sources and recommendation systems begin operating at the same time.
Other governments may eventually require AI providers to explain how political queries are handled, restrict candidate recommendations, create faster procedures for challenging synthetic campaign content or impose tighter rules immediately before voting.
For voters, the challenge is becoming more complicated than simply spotting a deepfake. They may also need to question where AI systems get their political information, whether those systems are making recommendations, and whether the person appearing on screen ever existed in that form at all. Other news: Instagram is cracking down on AI-generated influencers by requiring synthetic profiles to identify themselves or risk losing recommendation reach across surfaces such as Reels and Explore.