Financial systems, facial recognition, chatbot and scientific advances: Artificial Intelligence (IA) is present in all these solutions. Technology is increasingly incorporated into everyday processes, although, it is still synonymous with challenges. To regularize the use of technology, the European Union recently approved a historic agreement.
The regularization, signed between the European Parliament and the member states, includes standards that meet AI usability, social networks and search engines. After 37 hours of discussion, approval prohibits practices that threaten privacy and security or that are harmful at some level to the population. Now, The expectation is that the European agreement will serve as a facilitating example for the Brazilian reality.
Challenge for other countries
The new legislation puts the European Union ahead of the United States and Brazil itself in the “regulation race”. Nonetheless, Brazilian commissions were pioneers in debating the subject and AI applications even before the popularization of ChatGPT.
In February of 2020 the appropriate use of technology was already the subject of regularization debates. Although, establishing standards and restrictions becomes difficult as technology advances, making it difficult to launch legislation that favors all spheres involved.
Obligations and prohibitions of the agreement
At the moment, the balance of using AI is still being sought to avoid false news, blows, data breach and even discrimination against minority groups. That's why, the European agreement was made based on risk classification. To minimize these impacts, the agreement prohibits the use of real-time surveillance and biometric technologies can only be used by the police.
Furthermore, use may only occur in extreme cases, as terrorist threats, searches for people and investigations of serious crimes. The agreement also imposes specific obligations on services to disclose data used in training data.
Advances in AI also bring the need for new context analysis. The risk classification would have been directly impacted by recent models such as GPT-4, by OpenAI. For the European Union, the methodology would be included in the categories of greatest risk for users.
Legislation around the world
The European Commissioner, Thierry Breton, responsible for laws covering media and search engines, announced the agreement calling it a “historic milestone”. Nonetheless, the subject mobilizes authorities from around the world through the discussion of the appropriate use of Artificial Intelligence.
At the moment, already exist 24 countries that have some type of law to combat technology risks. But, China is the only country that has its own regulatory framework. Just like Brazil, the United States has not yet found a consensus on AI regularization.
