The European Union's groundbreaking Artificial Intelligence Act (AI Act) officially entered into force on Thursday, 1 August 2024, marking a significant milestone in the global regulation of AI technology.

As the world's first comprehensive AI regulation, EU lawmakers have argued that the AI Act aims to ensure the development and use of trustworthy AI systems within the bloc while safeguarding fundamental rights and fostering innovation.

The legislation introduces a risk-based approach to regulating AI applications, categorising them into four tiers: minimal risk, specific transparency risk, high risk, and unacceptable risk. Most AI systems, such as recommender systems and spam filters, fall under the minimal risk category and face no obligations.

However, high-risk AI applications, including those used in recruitment, loan assessments, and autonomous robots, will be subject to strict requirements such as risk mitigation systems, high-quality datasets, and human oversight.

The Act prohibits AI systems deemed to pose unacceptable risks, such as those manipulating human behaviour or allowing social scoring by governments. Certain uses of biometric systems, including emotion recognition in workplaces and real-time remote biometric identification in public spaces, are also banned, with narrow exceptions.

Margrethe Vestager, executive vice-president for a Europe Fit for the Digital Age, stated, "With the AI Act, the EU has taken an important step to ensure that AI technology uptake respects EU rules in Europe."

Generative AI, classified as "general-purpose" AI under the Act, will face specific requirements, including respect for EU copyright law, transparency disclosures on training data, and routine testing. However, the regulation provides some exceptions for open-source AI models, provided they meet certain criteria.

Enforcement of the AI Act will be overseen by the European AI Office, established in February 2024. Companies found in breach of the rules could face substantial fines of up to €35 million or 7 per cent of their global annual revenues, whichever is higher.

While the Act is now in force, most provisions will not take effect until 2026. Restrictions on general-purpose AI systems will begin in 12 months, and currently available generative AI systems like ChatGPT and Google's Gemini have a 36-month transition period to achieve compliance.

---