EU AI Act Cheat Sheet: Understanding Regulations in 2 Minutes

Ben Colman

Co-Founder and CEO

What is the EU AI Act?

The EU AI Act is legislation that regulates artificial intelligence based on risk levels. Think of it as the world's first comprehensive AI rulebook, designed to ensure AI development remains safe and trustworthy while fostering innovation.

How does it classify AI systems?

The Act divides AI into four risk categories. Unacceptable risk AI is completely banned, including social scoring systems and manipulative AI designed to exploit vulnerabilities. High-risk AI systems face strict regulations and must meet safety, transparency, and accuracy requirements. Limited risk AI requires lighter transparency obligations, while minimal risk AI remains largely unregulated.

What does the EU AI Act specifically say about deepfakes?

Deepfakes fall under the "limited risk" category, meaning they're not banned but must comply with transparency requirements. Starting August 2nd, 2025, developers and users must clearly inform people when they're interacting with AI-generated content, including deepfakes. This means deepfake videos, audio, or images must be clearly labeled so users know they're not authentic.

Some deepfake applications cross into prohibited territory. AI systems that use deceptive techniques to manipulate behavior and cause significant harm are banned outright. This creates a framework where legitimate deepfake uses remain legal with proper disclosure, while malicious deepfake applications designed to deceive or manipulate are prohibited.

When does this take effect?

Implementation follows a staggered timeline. Prohibited AI systems must comply within six months of the Act taking effect. General purpose AI models have twelve months, while high-risk systems have between 24-36 months depending on their classification. Crucially, deepfake labeling requirements become mandatory on August 2nd, 2025.

Why does this matter?

The EU AI Act sets global precedent for AI governance, influencing how companies worldwide develop and deploy AI technologies. Organizations operating in or serving EU markets must ensure compliance or face substantial fines up to 7% of global revenue.