European Union Reaches Initial Agreement On Artificial Intelligence Rules, Including Biometric Supervision

European Commissioner Thierry Breton, along with the formulators of the European Union's artificial intelligence bill. (photo: x @ThierryBreton)

JAKARTA - The European Union reached a temporary agreement on Friday December 8 regarding EU rules that include the use of artificial intelligence, including the use of the government in biometric surveillance and regulation of artificial intelligence systems such as ChatGPT.

With this political agreement, the European Union is heading to become the first major world force to enact laws governing artificial intelligence. Friday's agreement between European Union countries and members of the European Parliament comes after nearly 15 hours of negotiations following the nearly 24-hour debate the previous day.

Both sides will discuss details in the coming days, which could change the shape of the final legislation.

"Europe has placed itself as a pioneer, understanding the importance of its role as a global standard determinant. This is a historic day, I believe," European Commissioner Thierry Breton told a news conference.

This agreement requires basic models such as ChatGPT and generalized artificial intelligence systems (GPAI) comply with transparency obligations before being introduced to the market. This includes the preparation of technical documentation, complying with EU copyright law, and disseminating detailed summarys of the content used for training.

The basic model has a high impact with systemic risk of having to evaluate models, assess and address systemic risks, conduct adversarial tests, report to the European Commission on serious incidents, ensure cybersecurity, and report their energy efficiency.

GPAI with a systemic risk can rely on a code of ethics to comply with this new regulation.

The government can only use real-time biometric surveillance in public spaces in cases of certain crime victims, real threat prevention, attendance, or predictable, such as terrorist attacks, and searches for people suspected of committing the most serious crimes.

The agreement prohibits cognitive behavior manipulation, removal of targetless facial images from the internet or CCTV footage, social scoring, and a biometric categorification system to infer political, religious, philosophical, sexual orientation, and racial beliefs.

Consumers have the right to file complaints and receive meaningful explanations, while fines for violations will range from 7.5 million euros (IDR 126.1 billion) or 1.5% turnover to 35 million euros (IDR 588.3 billion) or 7% global turnover.

DigitalEurope's business group criticized the rule as an additional burden for the company, on top of other recent legislation.

"We have an agreement, but at what cost? We fully support a risk-based approach based on AI use, not the technology itself, but the last attempt to regulate the basic model has reversed this," said Director General Cecilia Bonefeld-Dahl, as quoted by VOI from Reuters.

The privacy rights group European Digital Rights is also critical. "It's hard to be enthusiastic about the law, which, for the first time in the EU, has taken steps to legalize public facial recognition directly across blocks," said Senior Policy Advisor Ella Jakubowska.

"Although Parliament is struggling to limit the damage, the overall package on biometric surveillance and profiling is the warmest," he said.

This regulation is expected to take effect at the beginning of next year after both parties officially ratify it and are expected to take effect two years after that.

Governments around the world are trying to balance the advantages of technology, which can have human-like conversations, answer questions, and write computer codes, with the need to set barriers.

Europe's ambitious AI rules come as companies like OpenAI, whose Microsoft is an investor, continue to find new uses for their technology, spark praise and concern. Google owner Alphabet, on Thursday 7 December, launched a new AI model, Gemini, to compete with OpenAI.

This EU law could be a blueprint for other governments and alternatives to China's lightweight US approach and interim rules.