A roundtable discussion on AI safety was hosted by UKTN in partnership with Shoosmiths and KPMG in September 2024.
- Participants discussed the potential regulation of AI, focusing on whether the UK should emulate the EU’s AI Act.
- Concerns about understanding and categorising AI risks were a central theme.
- There was debate on whether a sectoral approach to regulation is beneficial or if a broader framework is needed.
- The effectiveness of the EU AI Act and its impacts on innovation and business were scrutinised.
In September 2024, UKTN, in collaboration with Shoosmiths and KPMG, convened a roundtable to delve into AI safety and the potential need for regulatory measures. This discussion came in the wake of recent UK government summits addressing AI regulation. The deliberations operated under Chatham House rules, ensuring a discreet exchange of ideas among industry leaders.
A primary point of discussion was whether the UK should adopt a regulatory framework similar to the EU’s AI Act. The new UK government having prioritised AI regulation brought varied opinions on its implementation. Some advocated for a framework akin to the EU model, while others highlighted the unique needs of different sectors, suggesting a customised approach.
Addressing the notion that AI risks are not monolithic, the participants stressed the importance of categorising these risks. They identified both long-term existential threats and immediate risks, such as job displacement and intellectual property concerns, as pivotal areas needing policy intervention. Despite recognising the improbability of AI causing catastrophic outcomes, the ease of breaching current safeguards was alarming, necessitating a robust regulatory response.
The group deliberated on whether a sectoral regulatory approach, initially proposed by a previous government, should be maintained. While recognising the merit in addressing sector-specific needs, concerns about the varying pace of technological advancement in different industries were raised. Some participants argued for a comprehensive framework that could be quickly implemented to provide overarching guidelines, with industry-specific modifications developed as needed.
Attention was also directed towards the EU’s AI Act, exploring its implications for UK businesses in the absence of local legislation. There was a split in opinion on the act’s efficacy, particularly in balancing innovation with regulation. Concerns were raised about the impact of stringent rules on the release of transformative AI technologies within the EU, with some fearing that these could stifle creativity and technological advancement.
The discussion underscored the necessity for a nuanced, well-considered approach to AI regulation in the UK.
