The EU has, in recent years, strongly advocated the need to build a resilient Europe which is fit for the Digital Decade. This picture of Europe is one where people and businesses benefit from artificial intelligence-generated improvements (AI) in industry and in day-to-day life. However, also critical to this vision of the future, is an understanding that care must be taken to ensure that any AI-generated improvements are based on rules that safeguard the functioning of markets and the public sector, and also, people's safety and fundamental rights. For this vision to succeed, and for Europe to be internationally competitive, these values must be accepted at a global level and the EU must facilitate innovation across the Member States..
On 21 April 2021, in line with this vision, the European Commission (the Commission) published its AI package proposing new rules and actions which have, at their heart, an intention to turn Europe into the global hub for trustworthy Artificial Intelligence. The key elements were:
- An update to the 2018 EU Co-ordination Plan on AI.
- The introduction of the first ever legal framework on AI.
- The introduction of a new Machinery Regulation.
Co-ordination Plan on AI – 2021 update
Co-ordination of Member State efforts to develop, fund and implement AI is expected to strengthen Europe's leading position in promoting human-centric, sustainable, inclusive, secure and trustworthy AI systems. The initial 2018 plan defined actions and funding instruments for the development and adoption of AI which facilitated national AI strategies and provided EU funding for a variety of public/private AI initiatives. The 2021 update proposes concrete actions for collaboration between Member States to ensure that all efforts are aligned with the EU's strategy on AI and, with the Green Deal. Given the impact of the Coronavirus pandemic, it also has the objective of accelerating investment into AI with a view to boosting economic recovery.
Money to implement the updated plan will allocated from the Digital Europe programme, the Horizon Europe programme, the Recovery and Resilience facility and, the Cohesion Policy programme. Funds will be used to:
- Create enabling conditions for the development of AI and its widespread implementation.
- Foster AI excellence from 'lab to market' via a variety of public/private sector R&D and innovation initiatives.
- Ensure that AI is seen as trustworthy and working for the people by nurturing talent and promoting a vision of sustainable and trustworthy AI.
- Build strategic leadership in high-impact sectors and technologies including environment by focusing on AI's contribution to sustainable production, health by expanding the cross-border exchange of information, as well as the public sector, mobility, home affairs and agriculture, and Robotics.
New Regulatory Framework on AI
By putting in place a legal framework the Commission aims to ensure the protection of fundamental rights (e.g. the right to privacy, freedom of expression) and user safety. By addressing common concerns in these areas, it hopes to inspire trust in AI systems and promote their development and uptake.
The framework adopts a risk-based approach with AI systems being classified in one of four risk categories namely: Unacceptable, High, Limited and Minimal. As the heading suggests AI systems posing an unacceptable risk to European values and rights or to public safety will be banned outright. This would include, for example, systems utilizing subliminal messaging, those exploiting children and most biometric identification systems in public spaces for the purpose of law enforcement. High risk AI systems include those having the potential to create an adverse impact on people's safety or their fundamental rights (as protected by the EU Charter of Fundamental Rights). Thus, this category would include AI systems such as those linked to critical infrastructure, those involved in selection or examination processes, immigration systems, those linked to essential public and private services etc. AI systems such as this would be subject to tough regulation and would need to comply with strict obligations before they are put on the market. These would include:
- Demonstrating adequate risk assessment and risk mitigation systems
- The use of high-quality data sets to build the system in order to minimize risk and the possibility of discriminatory outcomes.
- Comprehensive logging of activity to allow for traceability of results.
- Detailed documentation of the system and its purposes.
- Clear and sufficient information for the end user.
- Appropriate human oversight measures.
- High level of robustness, security and accuracy.
Additional restrictions are also imposed on the use of all remote biometric identification systems.
Limited risk AI systems such as 'chat bots' have an obligation of transparency imposed on them; the end user must be aware that they are dealing with AI and not a person. Finally, minimal risk AI systems include things such as 'spam filters' and AI enabled games are free of restrictions but developers and promotors are encouraged to develop and follow a voluntary code of best practice.
The new Machinery Regulation will be complementary to the Regulatory Framework on AI. It is intended to ensure that all machinery products ranging from robots through to lawnmowers are safe for users and consumers. The AI Regulation will address the safety risks of AI systems ensuring safety functions in machinery, while the Machinery Regulation will ensure, where applicable, the safe integration of the AI system into the overall machinery, so as not to compromise the safety of the machinery as a whole. Businesses will need to perform only a single conformity assessment for both the AI and the Machinery Regulations. The new legislation will reduce manufacturers' administrative and financial burden by allowing digital formats for the instructions and the declaration of conformity, and by requesting an adaptation of fees for SMEs when a third party is needed for the machinery conformity assessment.
The way forward
The Commission's proposed Regulations are now subject to adoption by the European Parliament and Member States. They are not likely to meet with significant opposition and once adopted, the final Regulations will be directly applicable across the EU. In parallel the Commission will liaise with the Member States to move forward with the actions contained within the revised Co-ordinated Plan.
The content of this article is intended to provide a general guide to the subject matter. Specialist advice should be sought about your specific circumstances.