Summary of the points of the historic EU agreement on the regulation of artificial intelligence:
- Historical success:
- The European Union has reached a historic agreement to adopt the world’s first law regulating artificial intelligence (AI).
- Intensive negotiations:
- After four days of intensive negotiations, European representatives reached a consensus.
- Content of the law:
- The law sets restrictions on certain AI applications that are considered a threat to democracy and citizenship.
- It includes bans on biometric categorization systems, facial recognition, emotion recognition, social scoring and behavioral manipulation.
- Exceptions and conditions:
- Biometric surveillance systems are permitted in public spaces with a court order and for certain crimes such as kidnapping, human trafficking, sexual exploitation and terrorist threats.
- High risk:
- High-risk AI systems are defined in areas such as banking, insurance and the manipulation of election results.
- Transparency and documentation:
- Citizens can report violations and demand that AI comply with transparency requirements and European regulations.
- Basic models:
- Specific rules are set for basic AI models such as ChatGPT or DALL-E platforms.
- Sanctions:
- Depending on the size of the company, they provide for sanctions of 35 million euros or 7% of the global business volume.
- Criticism and concerns:
- The IT and communications industry criticizes the lack of in-depth debate and the possible slowdown in innovation in Europe.
- Come into effect:
- The agreement is the first step and could take around three years to enter into force.
- The legislation must be ratified by the European Parliament and the Council, which could delay final approval until the end of 2026.
- Global Impact:
- Despite the criticism, it is highlighted that two years after the first proposal, Europe has the world’s first legal framework regulating the development of artificial intelligence.