What happened
On 18 August 2026, TSE President Justice Kassio Nunes Marques convened the six other TSE ministers in a virtual/private plenary to rule on an AI-generated (deepfake) video of former president Jair Bolsonaro, recreated with his image and voice, shown at his son Senator Flávio Bolsonaro's presidential campaign launch. Votes cast across three related judgments (in the virtual plenary, reported by Brasil247/Poder360/G1) signaled majority support among ministers for a broad-veto interpretation: once content is characterized as a deepfake (synthetic recreation of a living person's image/voice used to help or harm a candidacy), it can be ordered removed regardless of whether it is proven capable of deceiving voters, and may constitute abuse of power even where AI use is disclosed. TSE had already ordered removal of a related anti-Lula video from the same camp on 14 August 2026.
Why it matters
This is a judicial/electoral-regulator interpretation of Brazil's 2026 election AI rules that narrows the space for disclosed AI-generated campaign content — moving from a 'disclosure suffices' standard toward a stricter per-se prohibition standard for deepfakes depicting real candidates/public figures, regardless of persuasive intent or disclosure. It directly shapes how political campaigns, AI content generation platforms, and social platforms operating in Brazil must treat synthetic media of real candidates during the active 2026 election cycle.
Action needed
AI content-generation platforms and campaigns operating in Brazil should treat any synthetic recreation of a real candidate's voice/likeness as high-risk regardless of disclosure; monitor for TSE's formal written resolution following the 18 August plenary votes for the binding standard.