EU Passes Landmark AI Act — 5 Things You Need to Know

The European Parliament has approved the AI Act, a milestone in the AI legislation landscape. With members of the European Parliament (MEPs) voting 523 in favor, 46 against, and 49 abstentions, the comprehensive act aims to protect democracy, fundamental rights, and environmental sustainability while fostering innovation.

The act includes bans on various AI applications that threaten citizens' rights, such as biometric categorization systems and the untargeted scraping of facial images for recognition databases. It also prohibits emotion recognition in workplaces and schools, social scoring, predictive policing solely based on profiling, and AI that exploits vulnerabilities.

According to Internal Market Committee co-rapporteur Brando Benifei (S&D, Italy), the Act will reduce risks, combat discrimination, and protect worker and citizen rights, emphasizing the centrality of human beings and European values in AI development.  

“Thanks to Parliament, unacceptable AI practices will be banned in Europe and the rights of workers and citizens will be protected. The AI Office will now be set up to support companies to start complying with the rules before they enter into force. We ensured that human beings and European values are at the very centre of AI’s development,” says Benifei.

Civil Liberties Committee co-rapporteur Dragos Tudorache (Renew, Romania) says: “The EU has delivered. We have linked the concept of artificial intelligence to the fundamental values that form the basis of our societies. However, much work lies ahead that goes beyond the AI Act itself”

Tudorache adds that AI will push for a reevaluation of the social contract at the core of democracies, education models, labor markets, and warfare conduct. The AI Act is seen as the initial step towards establishing a new governance model centered around technology. Attention must now be directed towards implementing this law.

5 Crucial Aspects of the Artificial Intelligence Act:

1. Banned use cases: The rules prohibit AI applications that endanger citizens’ rights, such as biometric categorization systems based on sensitive traits and indiscriminate scraping of facial images from the internet or CCTV footage for facial recognition databases. They also forbid emotion recognition in workplaces and schools, social scoring, predictive policing solely based on profiling, and AI that manipulates human behavior or exploits vulnerabilities.

2. Law enforcement exemptions: Law enforcement's use of biometric identification systems is generally banned, except in narrowly defined situations, such as targeted searches for missing persons or preventing terrorist attacks, with strict safeguards including time and geographic limitations and prior judicial or administrative authorization. Post-incident use ("post-remote RBI") requires judicial approval linked to a criminal offense.

3. Obligations for High-Risk Systems: High-risk AI systems, with significant potential harm to health, safety, rights, environment, democracy, and rule of law (e.g., critical infrastructure, healthcare, banking), must assess and mitigate risks, maintain use logs, ensure transparency and accuracy, and provide human oversight. Citizens can lodge complaints about AI systems and receive explanations for decisions affecting their rights.

4. Transparency requirements: General-purpose AI (GPAI) systems and their models must meet transparency requirements, including compliance with EU copyright law and publishing detailed training data summaries. More powerful GPAI models facing systemic risks must undergo model evaluations, systemic risk assessments, and incident reporting.

   Artificially or manipulated images, audio, or video content ("deepfakes") must be clearly labeled.

5. Supporting innovation and SMEs: Regulatory sandboxes and real-world testing, accessible to SMEs and startups, will be established at the national level to develop and train innovative AI before market placement.

The Act is undergoing final checks and is expected to be adopted before the legislature's end, pending formal endorsement by the Council. It will come into force twenty days after publication, with full applicability after 24 months, except for specific provisions with varying timelines.

The EU is among the first global blocs to establish comprehensive AI regulations, in contrast to the G7 and OECD, which have struggled to reach a consensus. This legislative initiative is considered a global benchmark for governments seeking to harness the benefits of AI while mitigating potential risks, such as disinformation, job displacement, and copyright infringement.

The European Commission proposed the legislation in 2021, facing delays due to disagreements over the regulation of language models accessing online data and AI usage by law enforcement and intelligence agencies.