EU Artificial Intelligence (AI) Law Negotiations Postponed Until December

Tudorache, EU AI legislator (photo: twitter @IoanDragosT)

JAKARTA - European Union policymakers have not agreed on a number of issues related to new artificial intelligence regulations ahead of an important meeting on Tuesday, October 24 which made an agreement impossible until December. This was said by four sources familiar with the matter.

The AI draft rule must be agreed upon by the European Parliament and member states of the European Union. To date, they have discussed it three times in the trilogy, namely the meeting between parliament and EU countries to formulate the final version of the law.

According to the source, a fourth trilogy meeting will be held on Tuesday, a day after EU legislators are scheduled to discuss their negotiating position around the basic model and high-risk AI systems.

The basic model, such as OpenAI's ChatGPT, is an AI system trained in large data sets, with the ability to learn from new data to perform various tasks.

Spain, which held the EU presidency until December, has pushed for an agreement and has proposed compromise in efforts to accelerate the process.

This includes a graded approach to set up a fundamental model, defined as those that have more than 45 million users, according to a draft seen by Reuters.

Spain also wants additional liabilities for a very capable fundamental model (VCFM), such as ChatGPT, including periodic checks to uncover potential vulnerabilities. Opponents say that smaller platforms can also have the same risks.

Spain said it had consulted other EU countries regarding potential compromise before the fourth trilogy. However, a final agreement was not possible at the meeting, the sources said.

The fifth trilogy is scheduled to be held in early December.

Failure to reach the agreement could push for negotiations until early next year. Discussions can then be further blocked by European parliamentary elections in June.

Several legislators, including EU industry chief Thierry Breton and co-rapporteur for the AI Act, Drago Tudorache and Brando Benifei, have expressed hope that the draft will be approved before the end of the year.

The EU will start working on the AI Bill Bill in 2021. In May this year, the European parliament agreed on draft regulations including new rules regarding the use of facial recognition, biometric surveillance, and other AI applications.

According to the proposal, AI tools will be classified based on the level of risk felt, from low to unacceptable. Governments and companies that use these tools will have different obligations, depending on the classification.