EU agrees historic deal on world’s first laws to regulate AI

The European Union (EU) has agreed on the details of the AI Act, a set of rules for the people building and using artificial intelligence (AI). EU policymakers hope that the framework is set to become the global benchmark for regulating AI.

The agreement was described as “historic” by Thierry Breton, the European commissioner who helped negotiate the deal. The deal reached in Brussels took three days of negotiations. Breton said 100 people had been in a room for almost three days to seal the deal.

“Europe has positioned itself as a pioneer, understanding the importance of its role as global standard setter,” Breton said

Carme Artigas, Spain’s secretary of state for AI, who facilitated the negotiations, said: “This is a historical achievement, and a huge milestone towards the future.” The agreement addresses a global challenge in a fast-evolving technological environment. She said that the provisional agreement “managed to keep an extremely delicate balance: boosting innovation and uptake of artificial intelligence across Europe whilst fully respecting the fundamental rights of our citizens.”

Artigas said France and Germany supported the text, amid reports that tech companies in those countries feared the regulation would impede innovation and their domestic businesses.

“The EU AI Act is a global first,” said European Commission president Ursula von der Leyen on X. It is “a unique legal framework for the development of AI you can trust. And for the safety and fundamental rights of people and businesses.”

Key Points of EU AI Act

By setting the new standards, the EU wants to pave the way for a global approach to artificial intelligence that is ethical, safe and trustworthy.

The provisional agreement can be simplified as follows:

• Rules on high-impact general-purpose AI models that may pose future systemic risks, along with regulations for high-risk AI systems.
• An updated governance system with some enforcement powers at the EU level.
• An expanded list of prohibitions, including the potential use of remote biometric identification by law enforcement in public spaces, with appropriate safeguards.
• Improved rights protection, requiring those deploying high-risk AI systems to conduct a fundamental rights impact assessment before putting the AI system into use.

The agreement on the EU’s AI regulation classifies AI systems into high-risk and prohibited practices. AI systems with limited risk face minimal transparency obligations, such as disclosing AI-generated content. Certain AI uses, such as cognitive behavioral manipulation or untargeted scraping of facial images from CCTV footage or the internet, are deemed unacceptable and will be banned in the EU. Similarly, emotion recognition in the workplace and educational institutions, social scoring, biometric categorization will also be banned.

Higher risk, stricter rules

The main objective of the EU’s AI act legislative initiative is to regulate AI based on the potential harm it could cause to society following a risk-based approach: the higher the risk, the stricter the rules. As the first legislative proposal of its kind in the world, the act could become a global standard for AI regulation in other jurisdictions just as the GDPR (general data protection regulation) has done for data privacy.

1. Minimal Risk (such as video games):
   Solutions falling into the minimal risk category, such as video games, will remain unregulated, ensuring continued availability as before.
2. Limited Risk (chatbots):
   Systems with limited risk, like chatbots, will be permitted but are required to meet transparency obligations. Users must be informed when they interact with AI.
3. High Risk (transport, recruitments, and loans):
   High-risk systems, involved in areas like transport, recruitments, and loans, must adhere to strict criteria before entering the EU market.
4. Unacceptable Risk (social scoring and facial recognition):
   Systems posing an unacceptable risk to people’s safety, livelihoods, and individual rights, such as social scoring and facial recognition, will face a ban.

What Next?

The final agreement is not yet public, and discussions are ongoing to finalize technical details. The agreements will be enacted following approval from both Parliament and the European Council, representing the 27 countries in the union.

Global response to AI regulations

The agreement puts the EU ahead of the United States, China and the UK in the race to regulate artificial intelligence and protect the public from potential risks.

The urgency to regulate AI heightened after the global impact of ChatGPT’s release last year, showcasing AI’s advancing capabilities. The Biden administration in the United States issued an executive order, emphasizing AI’s national security implications. Countries including Britain and Japan have varied approaches, while China has imposed some restrictions on data use and recommendation algorithms.