European Union: Practical Aspects Of Application Of EU Regulation On AI
- Companies should assess their current and future use of AI systems and list them in a model repository. Financial organizations can integrate AI into existing model management systems.
- AI systems are categorized by risk: unacceptable, high, limited, and minimal. Unacceptable systems, like real-time biometric identification, are prohibited, while high-risk systems, such as those in critical infrastructure or credit scoring, must meet stringent requirements and pass compliance assessments.
- Transparency is required for limited and minimal risk systems, ensuring users are aware of AI involvement. Compliance with the AI Act entails implementing ethical AI practices, particularly for high-impact General-Purpose AI (GPAI) models.
- To ensure compliance, stakeholders must assess AI risks, raise awareness, design ethical systems, assign responsibility, stay updated, and establish formal governance.
- Non-compliance with the AI Act incurs significant penalties, ranging from €7.5 million to €35 million or 1% to 7% of global annual turnover, depending on the severity of the violation. Understanding and adhering to the regulations is crucial to avoid penalties.
