EU lawmakers on Thursday will hold a key vote on whether to impose restrictions on the use of artificial intelligence such as ChatGPT in the European Union. European Parliament committees will outline their positions on negotiations with EU member states aimed at creating a law that will prevent abuses in the use of artificial intelligence and leave room for innovation.

Blok aspires to be a global pioneer in regulating technology that has aroused public and corporate interest over the past few months. Brussels’ movement towards this goal actually started two years ago with the proposal of the European Commission. EU member states submitted their negotiating positions at the end of last year.

But since then, the advent of ChatGPT, Midjourney, and other AI programs has led to the consideration of a number of changes, largely bringing the matter to the attention of Parliament. Following the vote in the committees on Thursday next month, the European Parliament will have a say in the plenary session.

“I think we propose a very good and balanced text that allows innovation while protecting the people,” said Brando Benifeuy, one of the main MPs for the text to be voted on Thursday.

double-edged sword

While AI holds great promise, it is also a double-edged sword as a technical tool. For example, it can save lives by improving medical assessments, or it can be used by authoritarian regimes to improve mass surveillance.

For the general public, the appearance of ChatGPT late last year was a source of curiosity and admiration; users were subscribing to watch him write essays, poems or translate in seconds.

While imaging AIs like Midjourney and DALL-E have sparked excitement online for creating the likes of Van Goghs or Pope in the puffer jacket, AI music sites have even impressed with their ability to reproduce human songs.

However, this technology has great fraud potential to deceive people and influence public opinion. This has prompted Elon Musk and some researchers to call for a moratorium until the legal framework catches up.

The position of the European Parliament corresponds to the main directives set out in the proposal of the commission, guided by the current EU law on product safety, which imposes inspection duties on manufacturers.

The core of the EU’s approach is to have a list of “high risk” activities for AI.

The Commission recommends that the designation cover systems in sensitive areas such as critical infrastructure, education, human resources, public order and migration management. Some of the proposed rules for this category will ensure human control over artificial intelligence, the provision of technical documentation, and the existence of a risk management system.

Each EU member state will have a supervisory authority to monitor compliance with the rules.

But many MEPs want to limit the “high risk” criteria to only cover AI applications believed to threaten safety, health or fundamental rights. Others, such as the green group, oppose it. Parliament is considering a number of specific obligations similar to those applied to the “high risk” list for manufacturer AI, such as ChatGPT.

Members of the European Parliament also want AI companies to protect against illegal content and copyrighted work that could be used to train their algorithms. The commission’s proposal requires users to be notified when they come into contact with the machine, and to indicate that the output of imaging applications is artificially produced.

Outright bans will be rare and only apply to practices that conflict with European values, such as the type of mass surveillance and citizen rating systems used in China. Lawmakers want to add bans to AI’s ability to recognize emotions and get rid of exemptions that would allow law enforcement to biometrically identify people remotely in public places. They also want to prevent photos posted online from being copied to train algorithms unless they get permission from the people involved.