Elaine Gemmell, head of regulatory affairs at InnoScot Health, analyses where the new EU AI Act legislation is beneficial and where it must remain flexible.
gopixa / shutterstock.com
1343083016
AI & Europe
The EU agreeing a landmark deal on regulation of artificial intelligence (AI) is a welcome development – but it must also remain open to change while ensuring global competitiveness.
While the proposed AI Act is yet to be ratified by member states and will not come into force until at least next year, it is set to be the world’s first comprehensive laws regulating artificial intelligence, governing the use of technologies such as facial recognition and ChatGPT.
EU Commissioner Thierry Breton has stated that the Act – agreed in December – will set “clear rules for the use of AI” while European Commission President Ursula von der Leyen insisted it is a "unique legal framework for the development of AI you can trust".
The Act will apply to EU AI providers and developers as well as those located in other territories – including the UK – if their AI systems affect individuals residing in the EU.
Nevertheless, I believe that the Act, which has been dubbed ‘Europe fit for the digital age’, must primarily help, not hamper the development of key technologies while ensuring safety and rights for the end user.
It’s good to see the EU stealing a march in this important, often contentious area, and the Act should serve to promote trust and investment thanks to greater clarity on how potential risks to health, safety and fundamental rights will be addressed.
That means the banning of single-purpose AI systems that pose unacceptable risks to individuals' fundamental rights, health, or safety, or to society. At the same time, so-called ‘high-risk AI systems’ will be scrutinised but can be permitted if their operators adhere to strict requirements before placement on the market or use.
However, with AI being a rapidly evolving area – including its role in the growth of important and innovative applications for healthcare – I believe that the new legislation must retain an element of fluidity and be ready to adapt to fresh developments as they arise.
The Act also must ensure the EU remains competitive and that AI companies aren’t discouraged from operating and innovating within it.
At this juncture, I would say it is vital for individuals and organisations that are likely to be affected by the AI Act to start looking at it in closer detail.
While the finer points of the Act, which organisations will need to fully assess for their potential strategic and operational impact, remain unknown for the time being, seeking appropriate advice now to prepare properly for it coming into force may well be prudent.
