The Lula Grok Hoax Turned an AI Chatbot Into a Smear Machine

Brazil's fact-checkers just mapped one of the fastest manufactured scandals of the election season, and it started with a cancer screening PSA. Days before the first round of voting on October 4, a line President Luiz Inácio Lula da Silva delivered about prostate exams was sliced out of context, run through an AI chatbot, and rebuilt as a pedophilia accusation. The Lula Grok hoax shows how the modern conspiracy supply chain works: a clipped video, a tricked chatbot, and an amplification network ready to pounce.

According to a Global Voices report drawn from the Brazilian fact-checking agency Aos Fatos, Lula was speaking at a campaign rally in the city of Belém when he talked about the importance of cancer prevention. His actual point was that many men resist getting prostate exams, unlike the culture of women's healthcare, where gynecological exams are routine from an early age. That line about women learning to be touched from a young age, meaning medical checkups, would become the visual evidence of the Lula Grok hoax. Aos Fatos obtained and checked one thousand eight hundred nineteen posts on X to reconstruct how the hoax was planted and harvested between the last two days of September.

How the Lula Grok hoax got its fake authority

Hours before the clipped video even surfaced, anonymous profiles in a far-right meme ecosystem were already experimenting with Grok, X's AI assistant. One user typed "cerveja e picanha," Portuguese for beer and filet steak, and asked the chatbot to summarize the expression with its initials in two letters, steering it to write "Lula consumes CP," an English acronym for child pornography. The chatbot pushed back, warning that the sentence was wordplay and that it had no records to support the claim, but the screenshot was the point. The Lula Grok hoax needed a source that looked like evidence, and an AI answer, even a debunked one, did the job.

The video side of the operation moved just as fast. According to Aos Fatos, Pam, an influencer who founded the Bolsonarista site The Incorrupt and belongs to the pro-Bolsonaro content hub Estudio 5, posted the first cut at eleven fifty-nine at night, reacting to Lula's line as if the president had said something unthinkable. Twenty-eight minutes past midnight, another account posted a second cut. Within a minute, posts were calling Lula "a pedophile" and asking users to pass the claim along. By the small hours of the morning, profiles were urging people to download and repost the video, pushing the hashtag #LULAPEDOPHILE, and one proposed picking out two sentences from the speech, building a hashtag around them, and mobilizing ten to twenty influencers. Forty-three percent of the posts analyzed, seven hundred forty-six of them, carried straightforward pedophilia accusations or encouraged spreading the claim. Every beat of the Lula Grok hoax was built for shareability.

How the Lula Grok hoax reached the trending tab

The narrative went mainstream when Nikolas Ferreira, a staunch ally of former president Jair Bolsonaro in the Chamber of Deputies, published one of the clips the next morning and followed it with a video responding to Lula's quote. The survey does not indicate that Ferreira took part in the Grok experiments or in creating the accusation, and he shows up as one of the first politicians with a wide reach to boost it, hours after it began circulating. The effect was immediate. A slogan reading "PROTECT THE CHILDREN" ranked twenty-eighth among X's trending topics just past nine, climbed to the top spot by quarter past eleven, while "LULA DON'T TOUCH THE CHILDREN" sat in fourth and "pedophile" in twentieth. For one morning, the Lula Grok hoax owned the platform's attention. Brazil's Superior Electoral Court gave Ferreira a twenty-four-hour deadline to remove his post the next day, with Minister Estela Aranha ruling that he had attributed an entirely different connotation to the original quote, dragging it out of its public health context.

Once the Lula Grok hoax was viral, it stopped needing the Belém speech at all. Accounts started reviving unrelated episodes and characters to cast Lula as part of an alleged abuse network. Some messages claimed he had defended or trivialized pedophilia, then drifted into talk of a "likely inclination" toward it. The fact that the rally took place in Pará state was exploited too: posts tied the story to documented sexual violence against children in the Marajó archipelago, even though no evidence links Lula to those crimes. Other posts recycled unproven claims made four years earlier by Damares Alves, then the minister for women and human rights, about children in Marajó, claims that were never registered by investigative bodies and that led Brazil's Federal Prosecution Office to file a civil action against her for spreading sensationalist false information.

The Lula Grok hoax kept absorbing older material. Some posts tried to tie Lula to Jeffrey Epstein through an email attributed to the disgraced financier, claiming the two had spoken by phone while Lula was imprisoned in Curitiba, with Noam Chomsky as the go-between. The email exists, but it proves nothing about whether the conversation happened. Valeria Chomsky denied that her husband had mediated any such call, the presidency's communications office said the call never happened, and experts told Folha de S. Paulo that it would have been incompatible with the rules of the federal police prison ward where Lula was held. The same email trail also mentioned former president Jair Bolsonaro, whom Epstein referred to once as "the man," according to the BBC.

The Lula Grok hoax is a blueprint worth studying because every step of it is reusable. A real sentence gets cut into a fake meaning. An AI chatbot gets prompted into producing a quotable screenshot. A network of accounts does the distribution work, and trending algorithms do the rest. The pattern is not unique to Brazil, which is why it keeps working everywhere, from manufactured bot-farm panic on the same platform to coordinated election conspiracies on X and the reflex that turns every tragedy into a plot. This time the lie got an assist from an AI, and the AI told them it was lying, and it spread anyway.