You probably encounter artificial intelligence (AI) in your daily life, whether you are unlocking your phone using facial recognition or using ChatGPT to find some information quickly. While AI technologies can benefit society, create jobs and increase productivity, their application could also compromise fundamental rights and jeopardise users’ safety. The European Parliament has worked to define rules for AI systems that strike the right balance between fostering investment in this new technology and protecting fundamental rights.
Since 2020, Parliament has adopted several resolutions outlining how the EU should regulate AI to support innovation, ethical standards and trust in AI technology. It launched a Special Committee on AI in a Digital Age. In May 2022, Parliament adopted its roadmap to AI. Parliament advocated a horizontal, innovation-friendly regulation framework, proportionate to the specific types of risk particular AI systems incur.
The European Commission’s 2021 proposed AI Act was the first of its kind in the world. Parliament formally adopted the law in March 2024, and it is expected to enter into force soon. The rules laid down in the act apply to all AI systems sold or used in the EU, to ensure that only safe products are placed on the market. As Parliament advocated, the proposal introduced a risk-based approach: certain AI practices with unacceptable, harmful risks will be prohibited, high-risk AI systems regulated, and transparency obligations will apply for systems with minimal risk.
Parliament secured important changes to the initial proposal. The definition of AI systems is now aligned with the Organisation for Economic Co-operation and Development (OECD) text. Moreover, the list of prohibited AI systems has been extended, and systems that influence voter behaviour are labelled high-risk. Parliament also ensured high-risk systems have to undergo a fundamental rights impact assessment before they are brought to market. Thanks to Parliament, citizens will be able to file complaints about AI systems, and receive explanations about decisions that affect their rights made using high-risk systems.
Parliament also succeeded in shaping the response to the rapid development of general-purpose AI (GPAI) models powering AI tools like ChatGPT. Characterised by their large size, opacity and the fact that they can be used and adapted beyond the purpose for which they were designed, these models present ethical and social risks: discrimination, misinformation and privacy violations. The AI Act introduces obligations on transparency and copyright law, and ensures content used for training for all GPAI models is disclosed. More stringent obligations will apply for more powerful, high-impact GPAI models. The newly named European AI Office, established within the European Commission, will have investigatory and enforcement powers over GPAI models, and a link to the scientific community to support its work.
Parliament acted swiftly in the face of the opportunities and challenges of this new technology. Parliament has set the agenda and shaped the new laws to ensure we can reap the benefits of new AI technologies while protecting our fundamental rights. Parliament’s powers fall broadly into six, often overlapping, domains: law-making, the budget, scrutiny of the executive, external relations, and, to a lesser extent, constitutional affairs and agenda-setting.
This graphic shows more examples of areas where Parliament used one or more of its different powers to influence legislation:
For a fuller picture of the European Parliament’s activity over the past five years, take a look at our publication Examples of Parliament’s impact: 2019 to 2024: Illustrating the powers of the European Parliament, from which this case is drawn.