Artificial Intelligence: Key Points of EU Law Coming into Force on August 1

On August 1, the European Union's first artificial intelligence (AI) law will come into effect, introducing regulations with specific transition periods. The law prohibits the use of AI systems that categorize individuals based on characteristics such as race, political or religious views, or sexual orientation.

It also bans the use of China's 'social credit' systems that aim to manipulate behavior and political loyalty through mass surveillance. Companies in the EU are prohibited from using AI to detect the emotions of their employees or influence people against their will.

The technology cannot be used to discriminate against individuals based on age, disability, or financial situation. Regarding facial recognition technology, law enforcement and security authorities can only use AI-powered facial recognition in public places with a court order.

The technology can be used to search for convicted or suspected criminals using recorded video footage, and in real-time for locating victims of human trafficking or preventing imminent terrorist threats. 'High-risk' applications of AI, such as those in law enforcement and personnel management, require human oversight, technical documentation, and risk management systems.

Developers must clearly label AI-generated content, and online platforms and national authorities are responsible for verifying this. The law carries significant fines for non-compliance, with penalties reaching up to millions of euros or a percentage of annual revenue.

The law will be implemented gradually, with various provisions coming into effect from 2025 to 2026. The EU also expects a 'AI Pact' of voluntary self-regulation from hundreds of companies by autumn 2023.