European Union rules on artificial intelligence are moving forward after reaching a “tentative” political agreement on the AI Act. As EU Commission President Ursula von der Leyen said, the law will help develop global standards, with the EU engaged internationally in organizations such as the G7, OECD, Council of Europe, G20 and the United Nations. This is, to date, the first official step in what tomorrow could be the world’s first sector-specific legislation. It was a necessary, courageous step on the part of Europe, but still burdened with uncertainties. And it is no coincidence that the official text is not yet there and will have to be voted on in the future.

AI Act between reliability and risk acceptability 

All enclosed in a framework of compromises reached held together by a sort of common thread that considers the reliability of AI in terms of the acceptability of risks, assessed mainly through the outcomes of compliance assessments rendered by the technology experts involved in the applications and governed by the accountability mechanisms dictated by the principle of accountability. The result has been a kind of regulatory fusion of accountability and acceptability that is complex and whose boundaries are still shifting instead as the technical side of the final text that the European bodies will vote on is currently being defined. Of course, before the vote, the completed text will have to go through the ritual of the legislative process and thus be approved by the representatives of the member states, confirmed by the EU institutions and only then formally adopted. 

The AI Act will become applicable, barring unforeseen events, two years after its enactment, with prohibitions and provisions that may already be operational after six months for some provisions. Save the possibility of voluntary compliance through the AI Act, allowing companies to comply before the regulation becomes formally operational. In the meantime, however, for many, the global race to establish technological leadership in artificial intelligence accompanied by the effort to develop “responsible AI” would now seem to be turning favourably toward the EU over China and the United States.

The most critical points

However, according to the European Consumers’ Organization (BEUC), this legislation has critical issues that will limit its effectiveness in protecting consumers. On the one hand, the AI Act prohibits social scoring, which is considered demeaning and discriminatory. It gives rights to consumers, such as the ability to file complaints against artificial intelligence systems or seek collective redress for mass damages. On the other, significant concerns remain: the law still allows artificial intelligence systems to identify and analyze consumer sentiments, which is considered invasive and inaccurate. The prevailing focus on high-risk systems leaves other areas, such as toys and virtual assistants, essentially unregulated. 

Widely used models, including ChatGPT, lack adequate regulations without independent third-party verification requirements or sufficient transparency for public scrutiny. In short, many crucial issues have been left under-regulated. It seems clear that the AI Act should have addressed consumer concerns related to manipulation and discrimination more thoroughly. Instead, it invokes other legislation, such as product safety, GDPR, and consumer law, to provide additional layers of protection. So, while the agreement on the AI Act is a significant step, critics equally stress the need for further improvements and rigorous enforcement to maximize consumer protection in the age of artificial intelligence.

Better to wait

Wouldn’t it have been better for the EU to put artificial intelligence legislation on hold until the technology is better understood? The risk is unintended consequences resulting from premature legislation that puts the race for regulation first instead of the race for innovation. Wanting to compensate through regulatory supremacy for lack of technological supremacy represents a daunting but probably necessary challenge for the EU to regain centrality in the geopolitical arena. And this is an obligatory path to avoid being trapped in the China-US tug-of-war while new competitors are already looming on the horizon.