started · updated
Brazil election integrity: Meta criticized over disinformation ads and TSE proposes deepfake rules FAST-MOVING
Amnesty International has warned that Meta’s advertising systems may be failing to protect the integrity of Brazil’s upcoming 2026 presidential election. In a recent experiment, the organization submitted 15 mock advertisements in Brazilian Portuguese to Facebook. The study found that eight of the submitted disinformation ads were accepted for publication, while seven others were flagged for identity verification rather than being rejected for disinformation.
The mock ads utilized three specific tactics: electoral delegitimization, enemy construction, and security-focused authoritarianism. These narratives included claims of electoral fraud and calls for military intervention. Amnesty International expressed concern that Meta is not doing enough to prevent its platform from being used as a vector for disinformation while continuing to profit from its advertising systems.
In a separate development, André Mendonça, a minister at the Tribunal Superior Eleitoral (TSE), has proposed a legal thesis to define deepfakes during election campaigns. He defines deepfakes as synthetic audio or video content that possesses a level of realism capable of creating a false perception of authenticity. This proposal follows a decision to remove an AI-generated video that falsely linked candidate Flávio Bolsonaro to businessman Daniel Vorcaro. The TSE is currently working to establish clear rules for the use of artificial intelligence in the electoral process.
Entities
Amnesty International · André Mendonça · Brazil · Facebook · Flávio Bolsonaro · Jurema Werneck · Meta · Tribunal Superior Eleitoral
Claims
What the coverage asserts, and how many sources carry each claim.
- [● 20 SOURCES] The mock ads included themes of electoral delegitimization, enemy construction, and security-focused authoritarianism. disa.org · portaldiarioderondonia.com · ccnnews.com.br · itapevinoticias.com.br · www.reporterpb.com.br · +15 more
- [● 20 SOURCES] Eight of the submitted disinformation ads were accepted for publication by Meta. www.amnesty.org · portaldiarioderondonia.com · ccnnews.com.br · itapevinoticias.com.br · www.reporterpb.com.br · +15 more
- [● 20 SOURCES] Amnesty International submitted 15 mock advertisements in Brazilian Portuguese to Facebook to test moderation. disa.org · www.amnesty.org · portaldiarioderondonia.com · ccnnews.com.br · itapevinoticias.com.br · +15 more
- [● 20 SOURCES] Meta's advertising systems may be failing to protect the integrity of Brazil's upcoming presidential election. disa.org · www.amnesty.org · portaldiarioderondonia.com · ccnnews.com.br · itapevinoticias.com.br · +15 more
- [● 10 SOURCES] The Brazilian presidential election is scheduled for 4 October 2026. disa.org · www.amnesty.org · portaldiarioderondonia.com · ccnnews.com.br · itapevinoticias.com.br · +5 more
- [● 20 SOURCES] Seven of the submitted ads were flagged for identity verification rather than being rejected for disinformation. disa.org · www.amnesty.org · portaldiarioderondonia.com · ccnnews.com.br · itapevinoticias.com.br · +15 more
- [● 8 SOURCES] André Mendonça ordered the removal of an AI-generated video linking Flávio Bolsonaro to Daniel Vorcaro. sbtnews.sbt.com.br · 98fmnatal.com.br · www.brasil247.com · oglobo.globo.com · www.jcatibaia.com.br · +3 more
- [● 9 SOURCES] André Mendonça proposed a criterion to define deepfakes as AI-generated content with a level of realism capable of creating a false perception of authenticity. sbtnews.sbt.com.br · 98fmnatal.com.br · cbn.globo.com · www.brasil247.com · oglobo.globo.com · +4 more
- [● 7 SOURCES] The TSE approved rules on the use of artificial intelligence during the general elections in October. www.jcatibaia.com.br · mauaagora.com.br · revistavigor.com.br · sbtnews.sbt.com.br · sitedaserra.com.br · +2 more