The National Data Protection Authority (ANPD) of Brazil recently rejected an appeal from Meta, the company that owns Facebook, seeking to use data from Brazilian users to train its Artificial Intelligence (AI). This decision highlights significant concerns about the ethics and security in using data for AI development.
Marcell Rosa, General Manager and Vice President of Sales in Latin America at Clevertap, warns about the risks associated with using data collected from platforms where misinformation and personal attacks are common. “When AI learns from biased and often harmful information, the risk is that these machines not only reproduce but amplify negative and distorted behaviors,” says Rosa.
The issue becomes worrying when considering the Brazilian electoral context. During the 2022 elections, the Superior Electoral Court (TSE) reported receiving over 500 daily alerts about fake news just in the second round.
The main concern is that AI, trained with opinions from ordinary people often influenced by misinformation, may replicate and amplify negative patterns. “AI tends to replicate human behavior, and when this behavior is characterized by polarization and lack of civility, technology inevitably reflects these flaws,” explains Rosa.
The ANPD’s decision is seen as an important step to ensure the responsible and ethical use of data in AI training. “Users’ privacy and information integrity are essential for the development of technologies that truly benefit society,” concludes Rosa.
This case highlights the urgent need for a more critical and regulated approach to data collection and use by digital platforms, aiming to protect the integrity of information and ensure that AI development is based on reliable and respectful data.
