Publications
We produce independent audits and public metrics to assess transparency, risk and bias of AI models in the Brazilian context. Here are the technical reports resulting from this work.
IDJÉ launches Espectro: Eleições 2026
IDJÉ • Salvador, 23 September 2026
Eleven days before the first round, IDJÉ launches the study “Espectro: Eleições 2026”, a Brazilian evaluation that measures the declared political leaning of language models.
A Student Does Not Grade Their Own Exam
Rodrigo Cavalcanti • September 2026
Public benchmarks can stop measuring capability once their items become training targets. This article examines how private tests, combined with open methodologies, reduce contamination and preserve the validity of independent evaluations.
Err or Withhold? Apparent Fairness Through Omission When LLMs Decide Under Stigma
Rodrigo Cavalcanti • July 2026
This work evaluates how seven artificial intelligence models each respond to 7,626 scored everyday situations (from a 7,653-prompt instrument with 27 neutral controls) involving people marked by social stigma. The analysis distinguishes prudence, systematic refusal and the reproduction of bias. The results show that apparently safe models may only be avoiding decisions, and that performance varies according to the type of stigma and the context presented.
The living and continuous version with more than 15 evaluated models is available in the Avalia section.