Recap of the European Parliament AI Proposal

1. The proposed regulation aims to create a legal framework for AI to ensure it is safe and respects EU laws and values (Chapter 1, Article 1).
2. The regulation applies to AI systems placed on the market, put into service, or used in the EU, regardless of whether the provider is established within the EU (Chapter 1, Article 2).
3. AI systems are categorized based on risk level, with high-risk AI systems subject to stricter requirements (Chapter 2, Article 6).
4. Providers of high-risk AI systems must ensure that their systems comply with various obligations, including data and data governance, documentation, transparency, and human oversight (Chapter 3, Articles 10-15).
5. Users of high-risk AI systems must ensure they use the system correctly and follow the instructions provided by the provider (Chapter 4, Article 29).
6. The regulation establishes a European Artificial Intelligence Board to advise and assist the European Commission and member states in AI-related matters (Chapter 5, Article 56).
7. National competent authorities will be responsible for monitoring and enforcing compliance with the regulation (Chapter 6, Article 59).
8. Penalties for non-compliance with the regulation include fines of up to 6% of the provider's total annual worldwide turnover or €30 million, whichever is higher (Chapter 7, Article 71).
9. The regulation encourages voluntary codes of conduct for providers and users of AI systems that are not considered high-risk to promote responsible AI practices (Chapter 8, Article 76).
10. The regulation promotes AI research and development by supporting a regulatory sandbox approach, allowing organizations to test AI systems in a controlled environment (Chapter 9, Article 77).

The European Artificial Intelligence Act classifies AI systems into four levels of risk: unacceptable, high, limited, and minimal. Here are descriptions and examples of each risk level:

1. Unacceptable risk: AI systems posing an unacceptable risk violate fundamental rights and are considered a clear threat to the safety or rights of individuals. These systems are banned or subject to strict limitations.
2. AI systems that manipulate human behavior, exploit vulnerabilities, or cause physical or psychological harm (Annex III, Point 1).
3. AI systems that use real-time remote biometric identification in publicly accessible spaces for law enforcement purposes unless certain exceptions apply (Annex III, Point 2).
4. AI systems evaluate the creditworthiness of individuals, leading to social scoring by public authorities (Annex III, Point 3).
5. AI systems enable governments to engage in indiscriminate surveillance, including the large-scale monitoring of individuals without their consent (Annex III, Point 4).
6. AI systems that use subliminal techniques to materially distort a person's behavior that could cause harm or is aimed at a vulnerable group (Annex III, Point 5).
7. High risk: High-risk AI systems are subject to strict legal and technical obligations, including data governance, documentation, transparency, and human oversight. These systems are typically used in critical sectors like healthcare, transportation, and law enforcement. Here are some examples:
8. Biometric identification and categorization of natural persons:
9. Management and operation of critical infrastructure:
10. Employment, workers management, and access to self-employment
11. Access to and enjoyment of essential private services and public services
12. Law enforcement
13. Administration of justice and democratic processes
14. Limited risk: AI systems with limited risk have some specific transparency obligations.
15. Example: Users should be informed when interacting with an AI system like a chatbot.
16. Minimal risk: Most AI systems fall into this category, and their use is encouraged for innovation and growth.
17. Example: AI-based recommendation systems used by e-commerce platforms or streaming services, suggesting products, movies, or music based on user preferences and behavior.

Please note that the European Artificial Intelligence Act does not explicitly define these four levels of risk. The Act primarily focuses on high-risk AI systems, while the other risk levels are derived from the regulatory requirements and obligations imposed on AI systems with different risk profiles.

There is much more to unpack and watch. As noteworthy items unfold, I will share those.

What are your concerns or questions about AI? Do you think the US should follow the same path?