New Guidelines for trustworthy Artificial Intelligence
High-level Expert Group publishes policy and investment recommendations.
SW – 07/2019
26 June 2019, a High-Level Expert Group set up by the European Commission
presented recommendations on policy and investment for trustworthy Artificial Intelligence (AI). The
experts made 33 recommendations that aim to steer a trustworthy AI towards
sustainability, growth, competitiveness and inclusion while strengthening,
promoting and protecting people.
experts generally recommend a risk-mitigation approach to regulating AIs,
taking into account proportionality and the principle of prevention of harm.
The greater the impact and/or the higher the probability of a risk caused by
AI, the stronger the regulation should be. The term ‘risk’ should be broadly
defined in order to cover all adverse effects both on the individual and on
for AI applications that present an unacceptable risk, the experts believe that
the principle of prevention of harm should apply, that is, regulators should
take precautionary measures if, for example, scientific evidence is available
about a risk to the environment or human health. Exactly what constitutes an
unacceptable risk should be discussed and decided after an open, transparent
and accountable societal debate, taking into account the European legal
framework, including the Charter of Fundamental Rights of the European Union.
importance of AI will also increase in the field of social security, whether
for medical care or the management of services and benefits, and will make it
necessary to examine opportunities and risks as well as the ethical and legal
published recommendations supplement the ‘Ethics Guidelines for
Trustworthy AI’ published in April and are
not claimed to be exhaustive. The experts only want to address the most urgent
areas for action. The guidelines contain policy and investment recommendations
for trustworthy AI, addressed to EU institutions and Member States. The expert
group believes that the EU institutions and the Member States have a key role
to play in achieving these objectives as key actors in the data economy, as
procurers of trustworthy AI systems and as benchmarks for sound governance.
its Communication ‘Artificial Intelligence for Europe’, the European Commission set out its
vision for trustworthy and human-centric AI. The aims of the Commission’s
- boost public and private investment
in AI in order to increase its uptake,
- prepare for socio-economic changes
resulting from AI,
- ensure an appropriate ethical
and legal framework to protect and strengthen European values.
Commission had already set up a High-Level Expert
Group on Artificial Intelligence for
this purpose. The Expert Group will support the implementation of the European Strategy
on Artificial Intelligence by developing ethics guidelines and strategy and
investment recommendations related to AI.
addition to the ethics guidelines published in April, the European Commission
has launched an assessment procedure to give all interested parties the opportunity to provide feedback on the
evaluation list which contains the most important requirements for trustworthy
AI. More than 300 stakeholders have already registered for the piloting process,
which runs until 1 December 2019. In early 2020, the Expert Group will review
the evaluation list on the basis of the results of the project and, if
necessary, propose further steps to the Commission.