European Union - AI Strategic Approach

Overview

The Communication "Fostering a European approach to Artificial Intelligence" (COM(2021)205 final), published on 21 April 2021, presents the EU's integrated strategy for AI based on two mutually reinforcing pillars: an "ecosystem of excellence" to boost research, innovation and market uptake, and an "ecosystem of trust" to ensure AI respects fundamental rights and safety. It accompanied the Commission's legislative package that included the proposal for a Regulation on AI (the Artificial Intelligence Act, COM(2021)206) and a revised Coordinated Plan on AI (2021). The Communication sets out enabling conditions (data access and sharing, computing capacity, investment and skills), regulatory principles (risk-based approach, human-centricity and rights protection), governance arrangements (coordination at EU and Member-State level) and international engagement priorities. The full text and annexes are published on EUR-Lex and the Publications Office (see COM(2021)205 (EUR-Lex) and Publications Office record).

Definitions

The Communication builds on previous EU AI policy framings (notably the 2018 and 2020 White Paper) and uses working definitions aligned with EU expert groups: "Artificial Intelligence" refers broadly to systems that display intelligent behaviour by analysing their environment and taking actions with some degree of autonomy. The Communication distinguishes between enabling conditions (data, compute, skills), deployment contexts (public sector, private sector, critical infrastructure) and risk classes (from minimal/limited risk uses through to high-risk systems that can harm safety or fundamental rights). It relies on the risk-based taxonomy later formalised in the AI Act proposal, where "high-risk" systems (for example in critical infrastructure, law enforcement, employment, education, healthcare and biometric identification) require specific conformity and oversight measures. The Communication also refers to concepts such as trustworthiness, explainability and human oversight as policy goals rather than prescriptive technical definitions.

Governance and Institutional Framework

The Communication calls for reinforced EU-level and national governance arrangements to accompany regulatory and programme measures. It proposes the setup of a European AI Board to help coordinate national competent authorities and ensure consistent implementation of EU rules (a function subsequently supported by Commission instruments and administrative decisions). It emphasises Member States' roles in implementing the Coordinated Plan and mobilising funding through Horizon Europe, Digital Europe and national recovery plans. The text asks the Commission to work with stakeholders, standardisation bodies and the European Data Protection Board to align AI rules with data protection obligations. Key institutional responsibilities remain with the Commission's Directorate-General for Communications Networks, Content and Technology (DG CONNECT) while enforcement and market surveillance functions are envisaged to be carried out by national authorities under EU coordination (see EC digital strategy library for related governance notes).

Key Focus Areas

The Communication identifies a set of focus areas to implement the twin objectives of excellence and trust. First, enabling conditions: the Communication stresses the need for access to large, representative, high-quality datasets, secure cross-border data spaces and increased computing capacity. It recommends public investment and incentives to mobilise private capital and urges Member States to coordinate investments via the Coordinated Plan on AI. Second, research and innovation: promoting research excellence hubs, European partnerships (data, AI and robotics) and an AI-on-demand platform to connect stakeholders. Third, testbeds and experimentation: creation of Testing and Experimentation Facilities (TEFs) and the strengthening of Digital Innovation Hubs and European Digital Innovation Hubs (EDIHs) to support SMEs and public administrations in adopting AI. Fourth, skills and education: upskilling, reskilling and curricula adaptation at all levels. Fifth, trust and regulation: designing a risk-based regulatory framework focused on high-risk systems, strengthened conformity assessment and market surveillance, and alignment with fundamental rights and data protection standards. Sixth, sectoral acceleration: focusing EU leadership in high-impact sectors (healthcare, transport, energy, manufacturing, agriculture, and public services), encouraging appropriate sandboxes and sector-specific coordination. These priorities are intended to be implemented through funding (Horizon Europe, Digital Europe) and regulatory instruments including the AI Act proposal.

Implementation Framework

Implementation envisaged by the Communication combines legislative, programmatic and coordination tools. The Commission proposed the AI Regulation (COM(2021)206) as the primary legal instrument for a risk-based approach; the Communication complements this with a revised Coordinated Plan (annexed) that sets concrete actions for Member States and the Commission to align investments, research agendas and public procurement. It calls for: (1) the development of common technical standards and certification schemes; (2) the establishment of testing environments (TEFs) and an AI-on-demand platform; (3) targeted funding and incentives for SMEs and startups; (4) integration of AI into public services and the uptake by administrations; and (5) alignment with other regulatory initiatives (GDPR, Data Governance Act, sectoral rules). The Communication defines sequencing and roles but leaves detailed operational rules to subsequent legal instruments, regulatory technical standards, and national authorities under an EU coordination mechanism.

Monitoring and Evaluation

The Communication requires monitoring through the Coordinated Plan and proposes indicators for tracking investments, skills, market uptake, research outputs and conformity frameworks. It envisages regular reporting by the Commission and Member States, periodic reviews of regulatory impacts, and coordination with existing EU monitoring tools (research and innovation scoreboard, digital economy metrics). The Communication also signals an iterative approach: the Commission commits to review and adapt measures as technology and markets evolve, relying on stakeholder consultations, the AI Alliance community and expert groups to inform updates and policy recalibration.

Penalties, Liability, and Appeals

As a Communication, the document does not itself create penalties but frames the enforcement approach later specified in proposed legislation (the AI Act proposal) and linked instruments. The Communication stresses the need for robust enforcement, market surveillance, and remedies for affected individuals, with national authorities responsible for supervision and sanctioning in conjunction with EU-level coordination. Liability and redress are addressed by aligning AI rules with existing liability regimes and reviewing whether additional civil liability clarifications are necessary; the Communication anticipates that conformity assessment and transparency obligations will underpin enforcement actions and administrative penalties in the regulatory instrument.

Relationship to Other Instruments

The Communication explicitly situates itself within the broader EU policy landscape: it builds on the 2020 White Paper on AI, the Coordinated Plan on AI, the GDPR, the Data Governance Act and related digital and cybersecurity strategies. It accompanies and supports the AI Act proposal (COM(2021)206) and references sectoral legislation (for example medical devices, product safety rules and transport). The Communication recommends coherence between new AI-specific rules and existing obligations, in particular data protection and consumer protection frameworks, and stresses coordination with standardisation bodies to operationalise technical requirements and conformity assessment processes.

International Alignment

The Communication commits the EU to active international engagement to shape global AI governance and standards. It promotes multilateral cooperation, support for interoperable standards through ISO/IEC and other bodies, and dialogues with partners to promote trustworthy AI principles. The Communication signals the EU's intent to protect fundamental rights in global fora while fostering cooperation on research, standardisation and trade. It highlights that EU regulatory choices aim to be interoperable with like-minded jurisdictions and adaptable to international developments.

Implementation Timeline

Sources and References