The idea of a politician who cannot legally run for office continuing to campaign anyway would once have been a contradiction in terms. In Brazil, artificial intelligence has quietly dissolved that contradiction. Jair Bolsonaro, the country’s firebrand former president, has been barred from standing for office until 2030 and is facing serious legal jeopardy, yet a digital version of him keeps turning up in front of supporters. According to The Australian, an AI avatar has become a tool for keeping the divisive figure present in Brazil’s political conversation, addressing his base even as the man himself is sidelined.
The context
Bolsonaro’s political troubles are well documented. After losing the 2022 election to Luiz Inacio Lula da Silva, he was ruled ineligible to run for public office until 2030 by Brazil’s top electoral court, and he has since been dragged into a sprawling legal case over an alleged plot to cling to power. That combination would normally push a leader out of the picture. Instead, his movement has leaned on a technology that was barely usable a couple of years ago: convincing, real-time synthetic likenesses that can speak, gesture and deliver a message in the leader’s own cadence.
The appeal is obvious. A physical candidate can be jailed, gagged by a court order or simply exhausted. A digital one cannot. It can be deployed to as many rallies, livestreams and social feeds as the operation can afford to run, and it never goes off script. For a movement built heavily around one personality, an AI stand-in offers a way to keep the flame burning while the succession question is sorted out behind the scenes.
The news
What makes the Brazilian example notable is not that a deepfake exists. Fake videos of politicians have been circulating for years. It is that the avatar is being used openly, as a deliberate campaign asset rather than a covert smear. The synthetic Bolsonaro is presented to supporters as a feature, a way of staying connected to a leader the courts have tried to remove from the field. That shifts the technology from the shadows of disinformation into the daylight of ordinary political marketing, and it blurs a line that electoral regulators around the world have not yet worked out how to police.
It also arrives against a backdrop of Brazil taking a comparatively hard line on AI in politics. The country’s electoral authority has already moved to restrict deepfakes in campaign material and to require disclosure when AI is used, part of a broader global scramble to write rules faster than the tools evolve. An avatar that is labelled and endorsed by the candidate’s own camp sits awkwardly inside those rules, neither a clear breach nor obviously permissible.
Two ways of reading it
Supporters and some technologists frame this as democratisation. If a leader has genuine backing, the argument runs, a digital presence simply lets that support express itself when the establishment has tried to silence him. On this view the avatar is closer to a very sophisticated broadcast than a fraud, because everyone watching knows the real man is not in the room.
The counter-argument is sharper. Critics see a dangerous normalising of synthetic politicians, where the emotional pull of a familiar face is manufactured on demand and detached from accountability. A human candidate can be questioned, cross-examined and held to what they said. An avatar answers to whoever writes its script, and the crowd’s loyalty is being harvested by an operation that may not share the leader’s judgment or restraint. Once voters grow comfortable cheering a computer-generated leader, the worry goes, it becomes far harder to insist that authenticity matters at all.
What it means for Australia
Australia is not Brazil, but the underlying problem travels well. Our own electoral cycle has already had a taste of synthetic content, and FluentSea has reported repeatedly on Australian Electoral Commission concerns about AI bots and foreign actors targeting political debate here. The Bolsonaro case sharpens the question because it is not about a hostile foreign operation planting a fake. It is about a domestic movement using an avatar of its own leader on purpose, which is a scenario our laws are even less prepared for.
Australia currently has no dedicated ban on AI-generated political advertising, and truth-in-political-advertising rules remain patchy across jurisdictions, with South Australia the notable exception. Disclosure obligations for AI content are still being debated rather than legislated at the federal level. That leaves an obvious gap. If an Australian party, or a well-funded campaign around a single popular figure, decided to run a labelled digital avatar at a by-election or a state poll, it is far from clear which rule it would break. Regulators would be reacting after the fact, as they usually are.
There is a commercial dimension too. The same voice-cloning and avatar tools driving enterprise adoption across Australian business are precisely what makes a campaign avatar cheap and convincing. The technology does not distinguish between a friendly customer-service bot and a political leader who cannot legally stand. As these systems become mainstream in Australian marketing and media, the temptation to point them at politics will grow, and the guardrails will need to be more than voluntary codes.
What’s next
The immediate test is whether Brazil’s courts and electoral authority treat the avatar as legitimate expression or as a circumvention of the ban that put Bolsonaro on the sidelines in the first place. That ruling, whenever it comes, will be studied closely by regulators elsewhere, because it will be one of the first real attempts to define where endorsed synthetic campaigning sits under the law. For Australia, the smarter move is not to wait for our own version of the problem to arrive before deciding what is allowed. The tools are already here, they are getting better every quarter, and the Brazilian experiment is a fairly loud warning about how quickly a banned candidate can become an unbanned pixel.
Sources: The Australian.
