The European Union is outlining broad plans to regulate artificial intelligence that could fundamentally reshape the way companies and governments use machine learning systems. The EU’s proposal aims to create a risk-based system for regulating AI, with strict rules for high-risk applications like self-driving cars or AI systems used to filter job candidates. Lower-risk AI systems like chatbots would face fewer requirements. The rules would prohibit systems that manipulate human behavior to circumvent users’ free will, exploit vulnerabilities for malicious purposes, or use social scoring by governments. Companies could face fines of up to 6% of global revenue for violations. The proposal is expected to be vetted by EU institutions over the next year or two before taking effect. It represents the first comprehensive effort by a major economic bloc to regulate AI and could set a global standard for the technology.