EU Resumes Negotiations on World's First AI Law

The negotiators from the European Parliament and EU member states began discussions on Wednesday. The meeting lasted for 22 hours.

After nearly 24 hours of negotiations, the EU failed to finalize a comprehensive law on artificial intelligence on Thursday but stated that the negotiations would continue on the following day.

Brussels aims to approve the world's first comprehensive AI law by the end of 2023. The urgency of this issue has increased since ChatGPT, a chatbot, demonstrated the rapid capabilities of AI and shocked the world last year.

ChatGPT's ability to generate fluent articles and poetry based on simple user prompts within seconds has amazed many people.

Despite the potential of the technology to change productivity and healthcare, critics point out that the risks of false information and abuse from AI, including deepfake images, are constantly rising.

The negotiators from the European Parliament and EU member states began discussions on Wednesday afternoon, with ambitious hopes to reach an agreement by early Thursday morning.

"There has been significant progress on the AI legislation in the past 22 hours. We will continue working with the European Parliament and Council tomorrow morning at 9:00. Stay tuned!" said Thierry Breton, the EU Internal Market Commissioner, on social media.

Although there is no real deadline, senior EU officials have invested a significant amount of political capital in completing the legal text by Thursday morning.

Even if the negotiators reach an agreement on Friday, the law will not come into effect until 2026 at the earliest.

There are two areas of disagreement between member states and the Parliament.

The first challenge is how to regulate foundational models used for various tasks. France, Germany, and Italy call for excluding these systems from the stricter parts of the law.

Some member states believe that regulations must limit the potential harm caused by AI misuse but still want to encourage innovation, especially as they aspire to have their own "OpenAI."

The second challenge is remote biometric surveillance, mostly through facial recognition using cameras in public places.

The European Parliament aims to establish a comprehensive ban on "real-time" remote biometric identification systems, but some member states want exceptions to allow the use of this technology for law enforcement.

The European Commission, the EU's executive body, first proposed an AI law in 2021, which regulates AI based on the level of risks they pose. For example, the greater the risks to civil rights or health, the greater the obligations of the systems.

Tech giants including Google and Meta are also seeking a place in the AI market.