Vestager has received less criticism from the U.S. regarding the AI Act compared to previous regulations, possibly due to the global implications of AI. The Act also addresses issues of bias in AI training data sets and requires transparency about the data's origins. However, the Act does not modify European copyright law, which could impact AI companies' ability to scrape data from the web for training their models. Despite concerns that the Act could disadvantage EU AI companies, Vestager argues that the regulations are necessary to build trust in technology.
Key takeaways:
- The European Union has agreed on the world’s first set of rules on the development and use of artificial intelligence, known as the AI Act. The Act uses a “risk-based” approach, applying different levels of regulation based on the potential danger of the AI system.
- The AI Act could potentially influence future actions adopted by U.S. lawmakers. European Commission Executive VP Margrethe Vestager, who leads the EU’s agenda on tech and digital issues, has been a key figure in the development of the Act.
- The Act includes measures to deal with bias in AI, requiring that the data AI algorithms are trained on reflect what they are doing and work without bias for specific situations. Certain use cases, such as state surveillance point systems and AI embedded in toys, are outright banned.
- There is ongoing debate about the impact of the AI Act on European AI companies, with some, including French President Emmanuel Macron, suggesting that the strict rules could disadvantage EU companies compared to their U.S. counterparts. However, Vestager argues that the Act is necessary to build trust in technology.