The European Union has developed the AI Act, a risk-based framework for regulating artificial intelligence, aiming to boost innovation and public trust. Proposed in April 2021, the Act ensures AI remains human-centered while providing clear rules for businesses. As AI use increases, its potential benefits come with risks, particularly where it affects individual rights. The Act intends to drive AI adoption, manage risks, and foster a local ecosystem. The AI Act differentiates AI applications by risk, excluding most uses from regulation. Military uses are also excluded since they fall under national jurisdiction. High-risk applications, such as those in critical infrastructure or law enforcement, require assessments before deployment and continuous compliance with standards like data quality and cybersecurity.

Medium-risk applications, like chatbots, face transparency requirements. Low-risk uses, such as social media content sorting, aren't regulated, though best practices are encouraged. The Act addresses general-purpose AI models (GPAIs), crucial to many applications, with particular attention to transparency and risk management. As AI technology evolves, especially with tools like ChatGPT bringing generative AI to the forefront, the EU's legislative approach has adapted. Negotiations concluded with the AI Act's political agreement in December 2023, with formal adoption in May 2024, setting compliance deadlines from 2025 to 2027. Enforcement varies by member state for most AI applications, while the EU oversees GPAIs specifically. Violations can lead to fines up to 7% of global turnover for serious offenses. The AI Act represents an ongoing effort to regulate AI as the technology and its societal impact continue to evolve.