Germany - AI Standardisation Roadmap

Overview

The German Standardization Roadmap on Artificial Intelligence (2nd edition) is a strategic roadmap prepared by the German Institute for Standardization (DIN) and the Deutsche Kommission Elektrotechnik, Elektronik und Informationstechnik (DKE) for the Federal Ministry for Economic Affairs and Climate Action (BMWK). It synthesises a multi‑stakeholder process with more than 570 experts and identifies over 100 concrete standardisation needs and six overarching recommendations for action. The Roadmap connects national standardisation priorities with the implementation of the EU Artificial Intelligence Act and provides a practical approach for developing harmonised technical standards, testing methodologies and conformity assessment procedures. The Roadmap is published in English and German and is available for download as the consolidated final report and annexes; the full text and complementary materials (flyer, slides, recordings) are available via DKE – AI Standardization Roadmap and the DIN AI pages at DIN – Standardization Roadmap AI. The document is intended to guide national and sectoral actions for standard development, testing and certification while supporting Germany's active participation in European and international standardisation fora.

Definitions

The Roadmap provides definitions and a working terminology for AI‑related concepts to promote consistent understanding across committees and sectors. Key definitions include 'AI system' (broad systems employing methods such as machine learning and statistical inference), 'high‑risk AI' (systems with potential significant impact on safety or fundamental rights, as informed by the EU AI Act), 'conformity assessment' (procedures for determining conformity with standardised requirements), 'data governance and data quality' (criteria for collecting, labelling and curating datasets used in modelling and validation), and 'sociotechnical system' (an AI system considered within its human, organisational and environmental context). The Roadmap emphasises consistent taxonomies and a common glossary to avoid fragmentation and support interoperability across committees and international standards bodies.

Governance and Institutional Framework

The Roadmap foregrounds a governance architecture that links DIN/DKE activities with a high‑level national coordination group mandated by the BMWK. This governance model coordinates working groups, assigns thematic priorities and tracks implementation through monitoring mechanisms. DIN and DKE act as national secretariats and convenors for technical committees and cross‑sector working groups; sectoral experts (industry, research, public authorities, civil society) are engaged through DIN.ONE and other collaboration platforms. The Roadmap also anticipates close collaboration with federal agencies such as the Federal Office for Information Security (BSI) and sectoral regulators to align technical standards with safety and cybersecurity requirements. For European alignment, the document references the role of the CEN/CENELEC JTC 21 and ISO/IEC JTC1/SC 42 structures in delivering harmonised standards supporting the AI Act. Governance responsibilities are distributed: DIN/DKE coordinate the technical programme; BMWK provides political steering and funding; public agencies contribute expertise and regulatory perspectives; market players implement and participate in standards development. See the public release and coordination statement at DIN – AI Roadmap.

Key Focus Areas

The Roadmap organises standardisation needs across nine core and cross‑cutting topics. Central focus areas include: (1) Foundations and terminologies — establishing a common taxonomy, semantics and AI glossary; (2) Security and safety — standards addressing safe operation, fail‑safe design, and certification criteria for safety‑critical AI; (3) Testing and certification — developing objective, reproducible test procedures and metrics for quality, robustness, accuracy and explainability; (4) Data quality and governance — metadata, provenance, labelling and privacy‑preserving techniques for training/validation datasets; (5) Sociotechnical systems — human‑centred design, human oversight and organisational processes; (6) Sector‑specific needs — medical AI, industrial automation, mobility, energy/environment and financial services each require tailored standards and testbeds; (7) Conformity assessment — horizontal and sectoral certification schemes and accreditation frameworks; (8) Cybersecurity and model security — resilience against adversarial attacks and supply‑chain risks; and (9) Interoperability and sustainability — data models, energy efficiency, and lifecycle assessment methodologies. The Roadmap also proposes pilot projects such as data infrastructure prototypes, a horizontal AI quality standard and dynamic modelling techniques. The document emphasises that harmonised standards are the technical basis for implementing legal requirements from the EU AI Act and for enabling trusted market access.

Implementation Framework

Implementation is framed as a phased, multi‑stakeholder programme. DIN and DKE prioritise work items and new technical committees based on urgency, regulatory milestones (EU AI Act timelines), market need, and international standardisation activities. The Roadmap recommends establishing pilot conformity assessment and certification programmes and accelerating standard development in priority clusters (risk management, dataset quality, transparency/logging, human oversight, robustness/cybersecurity). It calls for dedicated resources to support SMEs, funding for testbeds and data infrastructures, and collaborative tools (e.g. DIN.ONE) to broaden participation. The Roadmap envisages coordination of national inputs to European standardisation requests and active German leadership in CEN/CENELEC and ISO/IEC working groups. Implementation governance includes periodic progress reporting, public‑private pilot projects, and capacity building measures (training, guidance documents). For practical downloads and guidance on implementation steps, consult the Roadmap PDF at DKE/DIN – German Standardization Roadmap (PDF).

Monitoring and Evaluation

The Roadmap prescribes an ongoing monitoring approach to evaluate progress against prioritized standardisation needs. Monitoring is to be conducted via the Platform Learning Systems and the national high‑level coordination group, with periodic public reporting and key performance indicators (KPIs) such as number of standards initiated, drafts published, work items aligned with EU standardisation requests, pilot test outcomes, and stakeholder participation metrics. The Roadmap recommends establishing a living register of normative deliverables and an implementation tracker hosted by DIN/DKE to maintain transparency and enable course corrections. It also recommends commissioning independent evaluations and stakeholder consultations at defined milestones to ensure responsiveness to technological and regulatory changes.

Penalties, Liability, and Appeals

The Roadmap itself is a non‑binding standardisation strategy and does not create statutory penalties. However, it anticipates that harmonised standards will feed into conformity assessment schemes and regulatory requirements (notably the EU AI Act), which may trigger legal obligations and liabilities for providers of high‑risk AI systems. The document therefore addresses liability mitigation through improved documentation, traceability, testing regimes and certification—measures that reduce legal uncertainty and support defence against liability claims. It also recommends clear appeals and dispute resolution mechanisms within certification programmes, accreditation processes for notified bodies, and the alignment of national law and supervisory practices to provide legal clarity for market actors.

Relationship to Other Instruments

The Roadmap is explicitly designed to complement and feed into the EU AI Act and related European standardisation activities. It maps existing international standards (ISO/IEC SC 42, ISO, IEC, IEEE) and identifies gaps to be filled through national and European work items. The document refers to related national initiatives (German AI Strategy, AI competence centres) and sectoral roadmaps (e.g., Industry 4.0 norm roadmaps). It coordinates with cybersecurity and data protection frameworks (e.g., BSI guidance, GDPR compliance) and encourages coherence with safety and product regulations where AI is integrated. The Roadmap is positioned as a bridging instrument to translate legal obligations into technical specifications and test methods, supporting regulators, conformity assessors and industry actors to operationalise compliance.

International Alignment

A core objective of the Roadmap is to strengthen Germany's influence in international AI standardisation and to ensure interoperability of German solutions on global markets. It calls for proactive German engagement in CEN/CENELEC JTC 21, ISO/IEC JTC 1/SC 42, and related working groups, and it recommends prioritising deliverables that can be proposed as European or international standards. The Roadmap also highlights the need to align with neighbouring national bodies (e.g., BSI, British Standards Institution, AFNOR) and international standards efforts such as ISO 42001 and ongoing ISO/IEC activities. By aligning national work programmes with European and international roadmaps, the Roadmap aims to avoid fragmentation and to secure 'AI — Made in Germany' as a recognized label of trustworthy, standards‑based implementations. See European context in the Roadmap and CEN/CENELEC commentary at CEN‑CENELEC – DIN/DKE Roadmap note.

Implementation Timeline

Sources and References