In a ‘historic’ vote, the European Union became the first jurisdiction to adopt a landmark act on regulation of Artificial Intelligence (AI). The European Parliament legislative resolution of 13 March 2024 on the proposal for a Regulation of the European Parliament and of the Council of the EU on laying down harmonised rules on Artificial Intelligence (“Artificial Intelligence Act“) was adopted at the Parliament session in Strasbourg last week. 

GPAI (General-purpose) AI systems and their foundational models are mandated to fulfil specific transparency criteria, encompassing adherence to EU copyright law and the provision of comprehensive summaries of the training content. Enhanced GPAI models, with potential systemic risks, are subject to further prerequisites, including conducting model evaluations, assessing and mitigating systemic hazards, and incident reporting. Moreover, artificially created or manipulated multimedia content, known as “deepfakes”, must be explicitly labelled as such. This is to ensure clarity and maintain the integrity of information in the digital space. In addition, other high-risk AI systems will be placed under stricter scrutiny due to their potential detriment to human health, safety, fundamental rights, environment, democracy, and the rule of law.

In the spirit of safeguarding collective human rights, the EU AI act establishes guidelines that prohibit certain AI applications, such as invasive technologies with biometric categorisation, i.e. indiscriminate collection of facial images from the internet or CCTV for facial recognition databases. The rules also extend to emotion recognition in workplaces and schools, social scores, and predictive policing that relies on profiling an individual or assessing their characteristics. Moreover, any AI designed to manipulate human behaviour or exploit human vulnerabilities is prohibited.