Lupa study on election AI 920 posts; many unlabelled

Date unknown

The anchors describe a Lupa–TRE-RJ study identifying at least 920 election-related posts created or altered with AI between August 16 and September 28, averaging about twenty daily. Reporting live from TRE-RJ, Vladimir Platonow explains required synthetic-content labels and says 60% of cases lacked them. He describes fabricated reporting using recognizable journalists and falsely depicted celebrity endorsements, distinguishes deepfakes prohibited by electoral rules and warns about separating authentic from invented material. This is the study and its interpretation, with the broadcast's attributed legal explanation; it does not identify a newly enacted rule, every post as a deepfake or a proved effect on votes.

TV Brasil reports that a Lupa–TRE-RJ election-monitoring study identified at least 920 posts made or altered with AI between August 16 and September 28, about twenty a day, with 60% lacking required labels. Vladimir Platonow describes invented journalism using recognizable reporters and fabricated celebrity endorsements and explains the broadcast's distinction between labeling synthetic content and the prohibition of deepfakes. No effect on vote totals is established.