Further proposed changes to the regulation of artificial intelligence unveiled by European Commission

Posted: 14/06/2022

This article is written further to our article published on 19 May 2021, which discussed the European Commission’s proposal for regulation of artificial intelligence (AI) systems. The EU Commission unveiled changes to the proposed AI Act in November 2021. A summary of some of the key amendments is set out below.

The scope of the proposed regulation

The definition of an ‘AI system’ has been amended to clarify that it is a system that:

• Receives machine and/or human-based data and inputs.
• Infers how to achieve a given set of human-defined objectives using learning, reasoning or modelling implemented with the techniques and approaches listed in annex 1.
• These include: a) machine learning approaches, including supervised, unsupervised and reinforcement learning, using a wide variety of methods including deep learning; b) logic and knowledge-based approaches, including knowledge representation, inductive programming, knowledge bases, inference and deductive engines, reasoning and expert systems; and c) statistical approaches, Bayesian estimation, search and optimization methods.
• Generates outputs in the form of content (generative AI systems), predictions, recommendations or decisions, which influence the environments it interacts with.

The EU Commission has made clear that ‘general purpose AI systems’ (eg classic IT systems) do not, by themselves, fall within the scope of the regulations. This will, no doubt, be a relief to many, and also clarifies who will be considered a ‘provider’ where those systems are modified, integrated or rebranded and placed on the market.

AI systems exclusively developed or used for military purposes are excluded from the scope of the regulations, as are AI systems specifically developed and put into service for the sole purpose of scientific research and development.

Prohibited systems

The EU Commission has broadened the scope of this prohibition. It has done so by removing the limitation to ‘public authorities’ on the prohibition on AI systems used for social scoring, thereby extending the prohibition to private actors as well. It has also removed the limitation of ‘trustworthiness’ in respect of the prohibition to social scoring used to evaluate natural persons broadening the systems that will fall within this prohibition.

In addition, the definition of biometric data has been amended to include systems that do not ‘uniquely’ identify people, and the regulations will apply to biometric identification systems broadly, not only ‘remote’ systems.

High risk systems

The list of ‘high-risk AI systems’ has been expanded to include AI systems that:

• Are intended to be used to control or as safety components of digital infrastructure.
• Control emissions and pollution due to the serious incidents and irreversible damage to the environment and health that can be caused.
• Are used in insurance premium setting, underwriting and claims assessment which, if not duly designed, developed and used, can lead to serious consequences for people’s life, including financial exclusion and discrimination.

Obligations placed on users of high-risk AI systems have now been clarified to exclude those who use the AI systems for personal non-professional activity.

Conclusion

While it is unlikely that this is the last version of the draft regulations before their implementation, it does appear to show that the EU Commission has taken on board some of the concerns raised to date, and that legislation in this area will need to adapt quickly to keep up with this fast-paced area of innovation.

It is also interesting and encouraging that, as well as trustworthy AI being at the heart of the regulations, environmental factors are also at the forefront of the EU Commission’s intentions and are being incorporated in an important way into this legislation.  

This article has been co-written with Kate Abercromby, trainee solicitor in the commercial dispute resolution team. 

Return to news headlines