EU AI Act
欧盟于2024年正式生效的全球首个系统性AI监管框架，按风险等级将AI系统分类，高风险AI系统须满足严格的文档、日志、透明度和人工监督要求，违规最高罚款为全球年营业额的7%。
时间轴 (近 90 天)
EU AI Act provisions targeting General-Purpose AI (GPAI) models are now officially enforceable
Providers of general-purpose AI models must maintain detailed technical documentation explaining the model's training process, capability boundaries, and limitations, and provide it to regulators when required
Developers must publish a summary of training data usage, particularly regarding how copyright-protected content is handled
The EU AI Act imposes stricter requirements on models deemed to pose systemic risk
One criterion for determining systemic risk is the computational resources used for model training, measured in floating point operations (FLOPs) as a threshold
全部知识事实 (5)
Providers of general-purpose AI models must maintain detailed technical documentation explaining the model's training process, capability boundaries, and limitations, and provide it to regulators when required
90%待验证Developers must publish a summary of training data usage, particularly regarding how copyright-protected content is handled
90%待验证The EU AI Act imposes stricter requirements on models deemed to pose systemic risk
90%待验证One criterion for determining systemic risk is the computational resources used for model training, measured in floating point operations (FLOPs) as a threshold
90%待验证EU AI Act provisions targeting General-Purpose AI (GPAI) models are now officially enforceable
100%