European regulator hopes EU AI Act will set global example

The EU has taken a lead in setting safeguards around the new technology, and is now pushing to establish its framework as the regulatory benchmark, in a similar vein to the example set by the GDPR. The EU’s draft rules are expected to be approved before the end of the year. 

During the recent Reuters NEXT conference in New York, industry leaders came together to highlight the need to establish AI parameters to protect societies and democracies around the world.

In the US, discussions are ongoing in Congress regarding potential legislation to mitigate AI-related harms, including impacts on elections. The talks follow President Joe Biden’s signing of an executive order last month requiring developers of potentially risky AI systems to share safety test results with the US government before going live with new offerings.

Liz O’Sullivan, a member of the National AI Advisory Committee in the US, has emphasised the inherent dangers in AI, particularly the ways in which the technology draws information from learned situations and existing human biases. She spoke of proposed AI regulations, that might concern external audits, risk impact assessments, and the ability to turn off AI.

The US and China were two of 28 countries to join the UK in getting behind Rishi Sunak’s “Bletchley Declaration” recently – a pact that encourages formal collaboration between nations to build a safer AI future. In October, the G7 agreed on a voluntary code of conduct for AI development, setting a precedent for major countries in governing AI against a backdrop of anxieties over privacy and security risk. 

Benifei has noted alignment with global commitments but has also drawn attention to the importance of enforcing regulations through law, as opposed to commitments being voluntary. Such an approach would be needed, he argues, to address higher-level challenges in AI development, such as the risk of AI being weaponised.