EU AI Act Battlecard

The penalties within the AI Act vary based on the AI systems' risk level, the organization's size, and the exact violations by organizations. The penalty amounts can be as high as €35 million for specific breaches or 7% of the total global annual turnover, with lower penalties for other violations.

The AI Act's "risk-based approach" means that the obligation affects AI systems based on their level of risk posed. The AI Act categorizes all AI systems into four risk levels. Each risk level has its distinct obligations and provisions under the AI Act.

High-risk AI systems are subject to strict requirements such as comprehensive documentation, regular conformity assessments, robust data governance, and specific human oversight and transparency measures, such as the need for CE marking to denote their conformity.

Yes, AI applications and systems that pose significant risks to users' privacy, safety, and fundamental rights are outright prohibited. These include AI systems and mechanisms that manipulate human behavior without consent, exploit vulnerabilities of specific groups, enable social scoring, and deploy real-time remote biometric identification systems in publicly accessible spaces, with a few exceptions.