社会ガイド

EU AI Act Conformity Assessment and CE Marking

The AI Act’s conformity-assessment and CE-marking duties concern high-risk AI systems, not every AI product.

  • 3 分で読めます
  • 最終更新日
このページでは3 分で読めます
  1. 概要
  2. ディープダイブ
  3. 戦略的影響
  4. The Future of EU AI Act Conformity Assessment and CE Marking
  5. 現実世界の実装
  6. リスクとガードレール
  7. 実装ロードマップ
  8. 探検を続けましょう
  9. よくある質問

概要

The required route depends on whether the system falls under Annex III or is safety-related AI in regulated products, and on applicable standards, common specifications and product legislation.

ディープダイブ

Conformity assessment is the process of demonstrating that a high-risk AI system meets the Act’s requirements. The provider is responsible for classifying, documenting and assessing the system before placing it on the market or putting it into service. The AI Act does not require CE marking for all AI tools; the duties discussed here attach to high-risk AI systems, which may also be subject to other EU product rules. For Annex III point 1, Article 43 allows internal control or notified-body assessment when relevant standards or common specifications are applied. Annex VII is required if neither standards exist nor common specifications are available; the standard is not applied or only partly applied; available specifications are not applied; or an applied standard has a restriction, only for that restricted part. Points 2–8 use Annex VI internal control. For Annex I products, Article 43(3) integrates Section 2 and Article 17 quality-management requirements into the sector procedure. A sectoral no-third-party option conditioned on standards covering all sector requirements is allowed only if standards or Article 41 common specifications also cover all Section 2 requirements. After assessment, the provider draws up an EU declaration of conformity, assumes responsibility for conformity, and affixes CE marking visibly and indelibly, or to packaging or documentation if the system’s nature makes that appropriate. Digitally provided systems may use a digital mark if readily accessible. A notified-body identification number accompanies the CE mark when applicable. The CE mark signals conformity with applicable requirements; it is not a quality award or general guarantee. Substantial modifications can trigger a new assessment, while predetermined changes documented in the original assessment are treated differently.

戦略的影響

リスクと安全性

AI による壊滅的な被害も日常的な被害も、誰がリスクを理解し、誰が行動できるかにかかっています。

より明確な判決

国民と専門家のリテラシーは、強力な安全政策が政治的に可能かどうかを左右します。

誇大広告を打ち破る

明確な説明は、誇大広告、研究室の PR、曖昧な倫理劇場に囚われることを減らします。

The Future of EU AI Act Conformity Assessment and CE Marking

The legal framework anticipates harmonised standards, common specifications and notified bodies as parts of implementation, and Regulation 2026/1744 extended high-risk application dates. Technical standards and designation details may change, so providers should verify official Commission and EUR-Lex publications when selecting a route. Voluntary standards assist compliance but do not replace the regulation. Providers should monitor Commission publications on standards and conformity procedures before each launch. A change to intended purpose or system design can affect both classification and assessment, so maintain a controlled change log.

現実世界の実装

A provider classifies an Annex III system, completes the applicable conformity route, signs the EU declaration of conformity and affixes CE marking before market placement.

An Annex III point 1 system uses harmonised standards and chooses internal control or the specified notified-body route as the Act permits.

A high-risk system in an Annex I regulated product follows that sector’s conformity procedure with AI Act requirements integrated.

A software provider checks whether a digital CE mark is easily accessible from the system interface or through a machine-readable means.

リスクとガードレール

  • 能力が複雑になる一方で、実存的なリスクを SF として扱います。

  • 高度な自律性の下での調整による表面製品の安全性を混乱させる。

  • 英語以外や専門家ではない聴衆には、低品質の情報源しか提供されません。

実装ロードマップ

  1. 製品の危害、誤使用、制御不能/調整不良のリスクを分離します。

  2. どのような証拠がタイムラインと重大度についてのあなたの見方を変えるかを尋ねてください。

  3. マーケティング上の主張よりも、一次情報源と具体的な評価を優先します。

  4. 意識だけでなく、キャリア、政策、資金、スキルなど、行動経路を 1 つ特定します。

探検を続けましょう

Free newsletter

Get the daily AI briefing

Three verified AI stories every weekday morning, written in plain English. Free forever, no ads.

One email each weekday. Unsubscribe in one click. We never sell or share your address.

Test yourself

Take the EU AI Act Conformity Assessment and CE Marking quiz

Instant feedback on every answer, and a shareable certificate with a verifiable ID once you pass a course.

クイズを開始する

Support free AI education. AI Understanding is a 501(c)(3) nonprofit — no ads, no paywall, ever. Make a donation

よくある質問

What is EU AI Act Conformity Assessment and CE Marking?

The AI Act’s conformity-assessment and CE-marking duties concern high-risk AI systems, not every AI product. The required route depends on whether the system falls under Annex III or is safety-related AI in regulated products, and on applicable standards, common specifications and product legislation.

Which AI systems are the subject of the Act’s high-risk CE-marking obligations?

The conformity and CE provisions apply to high-risk AI systems, not every AI tool.

For Annex III points 2–8, which conformity route does Article 43 specify?

Article 43(2) specifies internal control for these categories.

When can Annex III point 1 providers choose between internal control and notified-body assessment?

Article 43(1) permits the choice where standards or common specifications have been applied.

When neither relevant harmonised standards nor common specifications are available, or a provider only partly applies a relevant harmonised standard, what route applies?

Article 43(1)(a) and (b) require Annex VII when both standards and common specifications are unavailable, or when a provider has not applied or has applied only part of a relevant harmonised standard.

How does conformity work for an AI system embedded in an Annex I regulated product?

Article 43(3) applies the sectoral conformity route and integrates the AI Act requirements.