While the news is full of cases of use of data from facial recognition by artificial intelligence – particularly in Asia – the European Union is considering the adoption of specific legislation aimed at regulating the uses of these technologies for further strengthen the protection of personal data of European citizens.
Awareness
In the resolution adopted on February 12, 2019 on “a comprehensive European industrial policy on artificial intelligence and robotics” (2018/2088(INI)), the European Parliament defines AI as a strategic technology of the XNUMXste century by emphasizing the imperative need to bridge the gap separating Europe from North America and Asia-Pacific and wishes to promote a coordinated approach at European level "to be able to compete with the massive investments made by third countries, notably the United States and China.
Highlighting the delay accumulated by Europe in terms of private investments in AI-related technologies (€2,4 to 3,2 billion in 2016, compared to €6,5 to 9,7 billion in Asia-Pacific and 12,1 to 18,6 billion € in North America), the European Parliament plans to meet this challenge by encouraging a global approach based on a strategic regulatory environment for AI while developing strong user protection.
Throughout this resolution, the European Parliament notes that “the existing schemes and doctrines can be easily applied” already forming a regulatory set capable of governing the operation of technologies linked to AI, governance and security of databases and the application layer – the real backbone of this technology – and personal data which are governed by specific provisions such as the GDPR.
However, the absence of a liability provision compromises legal certainty. The very nature of this technology inevitably makes civil provisions insufficient, or even obsolete.
European Parliament Resolution 2018/2088(INI) also aims for harmonization of national rules, noting that some countries have already legislated, which could lead to “excessive regulation in robotics and AI systems”.
The text also focuses on the issue of cybersecurity, another crucial aspect of artificial intelligence. A dual technology that can both combat and be subject to cyberattacks, Parliament considers it essential to prevent shortcomings and misuse of AI by applying “product security controls by market surveillance authorities and regulators.” consumer protection rules. To do this, it recommends that the European Union develop capacity and structural technological independence (Datacenter, systems and cloud component) as well as IT components, in particular processors. Laudable in spirit, this last point nevertheless seems to be an incantatory in view of the current situation.
The resolution therefore proposes to establish a legal framework based on the notion of ethics and promoting "ethic-by-design" technologies in a format close to the "privacy-by-design" model, developed by the General Regulations of data protection (GDPR).
Towards the regulatory initiative
In one article of August 22, 2019, Financial Times reveals that the European Commission is reportedly examining the possibility of imposing restrictive measures on the use of facial recognition surveillance technologies by private and public actors. In this case, the institution would consider drafting regulations – at community level – aimed at granting explicit rights regarding the use made of data resulting from the facial recognition of European citizens.
As a reminder, this leading institution at EU level, created by the Treaty of Rome in 1957 (articles 155 and 163), has the mission of legislative initiative and the implementation of community policies. While it plays a central role in guaranteeing the general interest within the Union, its members are nevertheless not drawn from universal suffrage. Made up of one European Commissioner per Member State (i.e. 27 Commissioners plus one for the United Kingdom), these are proposed by the heads of state or government of each country within the European Council and approved by the European Parliament.
According to the sources of the British economic and financial daily, this initiative would be part of the Commission's stated desire to support a legislative overhaul of the way in which the Union regulates the use made of technologies linked to artificial intelligence. , which should allow European citizens to “know when data [from facial recognition] is used, with the exception of any restrictions strictly supervised to ensure appropriate use [of these data]”.
While the Commission today enjoys the aura of pioneers after the adoption in April 2016 of the General Data Protection Regulation (GDPR or General Data Protection Regulation - GDPR), this now aims to create a framework (common to Member States) of ethical laws governing artificial intelligence (AI). According to an official interviewed by the Financial Times, the objective would be to “promote public confidence and acceptance” regarding new technologies “which carry specific risks [in terms of protection of personal data]”. The European Union seeks to reiterate the effects of the GDPR, establishing “a global standard for the regulation of AI [by setting] clear, predictable and uniform rules that adequately protect individuals”.
These initiatives reflect the fact that European institutions are striving more than ever to position themselves in the AI sector by adopting a unique approach, based on stakeholder involvement, so that in the future they can contribute effectively to the evolution of regulations.
First recommendations
These declarations follow a clear acceleration in the initiative taken by the European legislator on these subjects. In June 2018, the European Commission had already formed a group of high-level experts with the aim of supporting the implementation of the European strategy on artificial intelligence, in particular, through recommendations on the development of policies relating to the future, ethical, legal and societal issues of AI, including socio-economic challenges.
Made up of 52 independent experts from academic and industrial circles as well as civil society, the High-Level Expert Group on Artificial Intelligence (AI HLEG) published last June a first report specifying a series of ethical guidelines regarding artificial intelligence: legality (compliance with applicable laws and regulations), ethics (guarantee of absolute respect for ethical values) and robustness.
The AI HLEG offers an approach human-centric (human-centered) AI and lists recommendations for legislation and investment policies as well as a list of seven key requirements that systems must meet to be “trustworthy”:
- Human organization and supervision : Including fundamental rights, action and human supervision of AI;
- Security and technical robustness : Including attack resilience and security, protection provisions, accuracy, reliability and reproducibility;
- Privacy Protection and Data Governance : Including respect for privacy, data quality and integrity and access to data;
- Transparency : Including the traceability, exploitation and communication of the data collected;
- Diversity, non-discrimination and equity : Including the prevention of abusive bias, accessibility and universal design, as well as stakeholder participation;
- Well-being of society and the environment : Including viability, sustainability and respect for the environment as well as social, societal and democratic impact;
- Liability : Including auditing, minimization and reporting of negative impacts, arbitrations and appeals.
Although these recommendations and guidelines may seem relatively theoretical, the European Commission has made a particular effort in terms of policy applications and investments by formulating some 33 recommendations aimed at achieving a trust framework for AI.
This is indeed a necessary prerequisite, which will – ultimately – play a key role in the construction of the future European legislative framework for artificial intelligence.
Through this initiative, the EU and its institutions aim to encourage research and debate on the creation of an ethical framework governing this technology. Continuing a real lobbying action with multilateral bodies, in particular the OECD and the G20, the 27 wish to influence international work in order to be able to guarantee that the Union can remain competitive while benefiting as much as possible from the fallout from developments in these technologies.
The next step could be political. Succeeding Luxembourg's Jean-Claude Juncker at the head of the European Commission, Ursula Gertrud von der Leyen (CDU) will be inaugurated on 1er next November. The German has already declared that she will unveil – during the first 100 days of her mandate – legislation to provide “a coordinated European approach to the human and ethical implications of artificial intelligence”.
Jean Lebougre
