EU AI ACT – Why the EU is Leading the Way in Artificial Intelligence
December 14, 2023
Last Friday, the EU Parliament and European Council, which represents the 27 member states, reached a provisional political agreement on the Artificial Intelligence (AI) Act, so legislation can be voted on and come into force before the European election in June 2024. The text is yet to be finalised as technical details about how regulation will work in practice need to be discussed further and agreed on. However, we can offer some initial considerations.
A Landmark Legislation
In a nutshell the “regulation aims to ensure that fundamental rights, democracy, the rule of law and environmental sustainability are protected from high-risk AI, while boosting innovation and making Europe a leader in the field. The rules establish obligations for AI based on its potential risks and level of impact.”
Those who are familiar with the debate around AI and the EU Act know that this is a monumental achievement. The EU will lead the way on AI regulation, providing a risk-based, horizontal, human-centric framework other countries across the world, and the US in particular, are likely to look at while developing their own AI regulation.
What They Agreed On
During what is known in Brussels as ‘trialogue’, representatives of the EU Parliament, Commission and the 27 members strenuously negotiated over several outstanding issues, mainly on foundation models, governance, the scope of the legislation, and law enforcement.
- Free and open-source software, also known as a ‘general-purpose AI’ or ‘GPAI’ system, the foundation models that underpin ChatGPT for example, is exempt from the scope of the regulation unless they are categorised as a high-risk AI system, prohibited application, or an AI system at risk of causing manipulation.
- In terms of governance, the AI system’s compliance with the new regulation will be supervised by national competent authorities via the European Artificial Intelligence Board, which comprises member state representatives, alongside a new European AI Office, which will sit within the EU Commission. The new AI Office will also supervise the implementation and enforcement of the new rules on general-purpose AI models.
- Member states managed to secure an exemption on the use of biometric identification systems (RBI) in publicly accessible spaces for law enforcement purposes. This means that biometric categorisation systems based on sensitive personal traits like race, political opinions and religious beliefs are banned unless those characteristics have a direct link with a specific crime or threat.
Sanctions and Fines
Depending on the company size and violation, fines for non-compliance range from 7.5 million euros or 1.5% of turnover for the supply of incorrect information to 35 million euros or 7% of global turnover.
Meanwhile, the US and UK…
As mentioned, the EU AI Act is the world’s first example of AI regulation, and it will inevitably become the benchmark for forthcoming regulation in other countries. In October, the White House issued an Executive Order (EO) that establishes, “new standards for AI safety and security, protects Americans’ privacy, advances equity and civil rights, stands up for consumers and workers, promotes innovation and competition, advances American leadership around the world, and more.” The US approach differs from the EU approach broadly in two aspects: legal obligations and fines. Whilst the EU Act creates new legal obligations for AI providers and fines for those companies that don’t comply, the EO adopts a rule-based approach and does not contain enforcement provisions.
The announcement was made in conjunction with the UK AI Safety Summit where world leaders gathered to discuss a shared approach to AI risks and opportunities. Although the Bletchley Declaration is undoubtedly a success story for this Government, it is unlikely that the UK, despite being one of the leading centres for AI, could be the leading force in AI development, if compared to the US and China.
At this point, the co-legislators (European Parliament and Council) will vote on the final text, which is expected in early 2024. All obligations are to be complied with within two years after the text enters into force, with certain provisions becoming enforceable from as early as six months after the text enters into force.