European Commission launch AI legislative proposal
This week the European Commission has launched its legislative proposal on AI, outlining how AI could be developed and used in Europe in the future. As expected, the legislative proposal is extensive, so here are four key announcements to be aware of:
- Prohibited practices
The Commission has taken a bold step towards prohibiting a number of AI practices (see Article 5).
The prohibitions cover practices that have a significant potential to manipulate someone beyond their consciousness or exploit vulnerabilities of specific vulnerable groups such as children or people with disabilities, in order to distort their behaviour in a way that may cause them harm. The proposal also prohibits AI-based social scoring by public authorities, such as that already deployed in much of China. The use of ‘real time’ remote biometric identification systems in public spaces for the purpose of law enforcement is also prohibited but with limited exceptions, such as using the technology to find missing children or in the threat of a terror attack.
Currently most of the listed prohibited applications are open to interpretation. For example, what practices will be considered manipulative or harmful? The devil will be in the detail and it will come down to the final wording of the bill’s articles to determine what will and won’t fall in scope.
Non-compliance with the prohibition of these AI practices could have a devastating impact on an organisation's finances with potential fines of up to 30 000 000 EUR or, up to 6 % of the companies total worldwide annual turnover, whichever is higher.
- Defining high risk applications
The current legislative proposal attempts the difficult task of establishing a definition of high-risk AI. These uses are identified by the Commission as having a high likelihood and potential to cause severe harm to an individual or wider society.
Using this definition, the Commission will maintain and update an Annex of high risk AI applications. The current proposed list includes use in transport, education, recruitment, credit scoring and law enforcement. (Full list available in Annex three). This list may of course be expanded or reduced during the legislative process.
- New European Artificial Intelligence Board
A European Artificial Intelligence Board (EAIB) will be created to oversee the implementation of these rules. The Board will compromise of representatives from the Member States and the Commission. This board will have a huge amount of influence over what gets added to the high-risk list and prohibitions list, meaning exactly who sits on the board will be key.
- Conformity assessment for high risk applications
Providers of high-risk AI will have to satisfy a set of ex-ante requirements before deploying their product or service within the EU, including around the quality of the training data set and the level of human oversight. The current list of obligations for high-risk systems can be found under Article 16. This list of requirements will inevitably face a lot of scrutiny over time and be subject to change.
Commenting on the launch, Sue Daley, Director of Tech and Innovation, said:
The European Commission’s recent announcement of their legislative proposal for AI will be monumental in shaping Europe’s AI future and will have signification implications for the development and adoption of AI applications in Europe. It will take some time to understand the full impact of this proposal, but it will undeniably have broader consequences for the entire global AI industry.
As the legislative process now moves over to the European Parliament, we call on legislators to work together with industry and like-minded international partners, to find solutions to some of our common AI challenges and strike a balance between regulation and innovation that ultimately promotes trust in and drives forward the adoption of ethical and responsible AI.
Over the upcoming months and years this proposal will continue to be shaped and amended as it moves through the legislative process. If you’d like to find out more about the European Commission’s AI proposal, what it could mean for industry and next steps for engagement, make sure to register here for our next AI Leaders Lunchtime Lecture with Herbert Swaniker at Clifford Chance.