25 May 2023

Event round-up: How best to prepare for the EU’s AI Act

As the EU’s AI Act continues to progress through the European Parliament, techUK convened industry leaders to understand what the latest developments in this area could mean for our members and how the industry can best prepare for what’s to come. The panel was specifically encouraged to explore the differences between the EU and UK approaches to AI governance, examine international interoperability and share best practices for industry preparation

techUK's Head of Data Analytics, AI and Digital ID Katherine Holden was delighted to be joined by a fantastic panel of legal and industry experts: 

  • Sarah Cameron (Legal Director, Pinsent Masons) 

  • Alex Kirkhope (Partner, Technology and Outsourcing | AI, Shoosmiths) 

  • Jaspreet Takhar (Senior Associate, Healthcare | Data | AI & Tech, Baker McKenzie)  

  • Ryan Donnelly (Founder | Enzai – AI Governance) 

The conversation started by examining the origins and the approach of the EU AI Act, which is risk-based, based on ethical principles and fundamental rights. The Act aims to establish the EU as a global leader in AI regulation and adopts a uniform approach with broad territorial scope. While the Act has faced criticism for its horizontal, risk-based, tech-focused approach (deemed by some as too prescriptive), it demonstrates the EU's commitment to being at the forefront of AI governance. The panel also discussed the potential influence of the market on the Act's global impact, similar to the "Brussels effect”. 

The contrasting regulatory approaches between the EU and the UK was the main topic during our webinar. The panel explored how the UK aims to foster innovation and entrepreneurship by embracing a flexible and proportionate approach to AI regulation. The UK Government’s focus is on establishing principles and frameworks that encourage responsible AI development while promoting adaptability and facilitating growth. On the other hand, the EU has taken a more prescriptive and detailed approach, akin to product-safety regulation. Through the EU AI Act, they impose a comprehensive set of technical and organizational requirements on developers and users of high-risk AI systems to ensure accountability and address potential risks. The sustainability of the UK's iterative and flexible approach was a topic of discussion, considering the growing demands for regulation from tech leaders, and the US Government's consideration of an AI Bill of Rights. Understanding these differences and their implications for businesses operating in both the EU and the UK was emphasised as crucial by the panel and participants. 

The webinar also explored the changes introduced by the EU AI Act and the subsequent criticisms it has faced. The Act has been updated and now includes higher penalties, aligns with OECD definitions, and places emphasis on machine learning and deep learning. The panel mentioned that it now also addresses individual rights, such as explanations and representative actions, and covers generative AI and foundational models. However, concerns were raised about blind spots and the need for consumer rights protection. The EU's perceived prescriptive early approach, despite attempts to incorporate a more holistic perspective, was highlighted as a challenge. 

Interoperability and international alignment were other key themes discussed during the webinar. Speakers identified collaborative efforts and alignment between the EU and the UK as desired outcomes to foster innovation and trusted adoption. They acknowledged that establishing common standards, definitions, and frameworks is crucial for achieving interoperability. Potential alignment exists in areas such as foundation models and generative AI, particularly when specific risks are involved. The upcoming 2024 elections in the UK and EU were identified as significant factors influencing the potential political alignment of the two regimes. Lastly, the panel advocated for larger global input and cooperation, specifically engagement with countries beyond the EU, UK and US. 

The event concluded with a focus on best practices for industry preparation. Speakers recommended and highlighted the following points: 

  • The importance of considering compliance and risk management systems from the outset was emphasised. 

  • Engaging with industry bodies, regulators, and AI standards hubs was encouraged to stay up to date with the evolving landscape. 

  • Establishing effective governance structures, fostering alignment between stakeholders, and avoiding duplication of efforts were identified as crucial steps. 

  • Existing laws, such as intellectual property (IP) laws and the General Data Protection Regulation (GDPR), should not be overlooked, as the AI Act builds upon these established principles. 

Overall, the event provided insights into the EU's AI Act, the contrasting approaches to AI governance between the EU and the UK, the changes introduced by the Act, the need for interoperability and international alignment, and best practices for industry preparation. 

Please also see below for other events in the pipeline! 

  • Attend our in-person launch of the DSIT AI Assurance repository techniques on June 7. RSVP here. 

  • For techUK members, join the first meeting of our generative AI working group on June 20. RSVP here. 

Amelie Soares

Amelie Soares

Tech & Innovation Intern, techUK