The regulation, agreed in negotiations with member states in December 2023, was endorsed by MEPs with 523 votes in favour, 46 against and 49 abstentions.

It is still subject to a final lawyer-linguist check and needs to be formally endorsed by the Council of the European Union.

Welcoming the progress, Barry Scannell, a consultant at William Fry specialising in AI law, told Irish Legal News: “The passing of the AI Act by the EU Parliament is such a big deal, with only the formality of the linguistic approval left.

“Following today’s vote, the AI Act should take effect before the summer, with the first provisions, banning certain types of AI systems, likely being in place before the end of the year.”

He continued: “The Act bans certain AI practices deemed manipulative, exploitative, or invasive, including deceptive AI, exploitation based on vulnerabilities, biometric categorisation of sensitive information, social scoring, real-time biometric identification in public spaces, criminal offence risk assessment, creation of facial recognition databases through untargeted scraping, and emotion inference in workplaces and educational settings.

“Over the next couple of years, further provisions will enter into force relating to general purpose AI models behind the likes of Gemini or ChatGPT, as well as regulations on high risk AI systems.

“Violations carry hefty fines up to €35 million or seven per cent of global turnover.

“I think this is a great development as it protects EU citizens from the downsides of this incredible new technology, while providing companies with a framework around which they can innovate, use and develop new AI systems. A drawback will likely be high compliance costs and the complexity of the requirements, particularly for SMEs, but overall I think this is a very positive development.”