Updated 13 May 2026 · The Digital Omnibus AI Act (trilogue agreement 7 May 2026) preserved Article 6(3). The mechanism now converges towards Annex VIII Section B: non-high-risk self-classification becomes a public artefact with documented justification deposited in the EU database. The Sprinkling Act report produces exactly this type of methodological documentation — without substituting for legal advice.
Regulatory angle — Article 6(3)
Were you correctly classified as high-risk?
Article 6(3) of the EU AI Act allows an Annex III system to exit high-risk classification — under documented conditions. The exception exists in the text. The defensible documentation rarely does.
Most companies in Annex III sectors accepted high-risk classification by default. Few have formally tested Article 6(3) — the derogation that allows an Annex III system to exit high-risk where the provider demonstrates it does not pose a significant risk of harm. Post-Digital Omnibus (trilogue agreement 7 May 2026), the derogation was preserved by Parliament and Council against the Commission's proposal to remove it. But the mechanism changed: under the agreed Annex VIII Section B, providers self-classifying non-high-risk must register the system in the EU database with documented justification — even when they consider their system exempt from high-risk classification.
- · Methodological assessment — not legal advice
- · Article-by-article: Article 6(3) + Annex III §1–§8 + Annex VIII Section B + Article 49
- · Documented justification deposable as a public artefact (post-Omnibus 7 May 2026)
Classification
When Article 6(3) becomes a path
Article 6(3) of Regulation (EU) 2024/1689 provides that a system listed in Annex III shall not be considered high-risk where it does not pose a significant risk of harm to the health, safety or fundamental rights of natural persons, including by not materially influencing the outcome of decision-making. The derogation applies where ONE of four conditions is met. A profiling clause carve-out remains: a system that performs profiling of natural persons is always high-risk, regardless of the four conditions.
Test: does your AI system fit one of the four conditions below AND avoid the profiling carve-out? If yes — Article 6(3) becomes a potentially eligible path, subject to documented self-assessment and public registration under Annex VIII Section B.
| Condition | Article 6(3) | Annex VIII Section B |
|---|---|---|
| (a) narrow procedural task | Potentially eligible | Documented justification required |
| (b) improvement of prior human activity | Potentially eligible | Documented justification required |
| (c) detection of patterns without replacing human review | Potentially eligible | Documented justification required |
| (d) preparatory task to an Annex III assessment | Potentially eligible | Documented justification required |
| Profiling of natural persons | Always high-risk (carve-out) | Full Annex III registration required |
Source: Reg. (EU) 2024/1689 Article 6(3) verbatim · profiling carve-out in last subparagraph of Article 6(3).
Timeline
Article 6(3) post-Omnibus — the derogation preserved
The Digital Omnibus on AI (trilogue agreement 7 May 2026) preserved Article 6(3). The European Parliament and Council rejected the Commission's proposal to remove the derogation. The exception remains accessible. But the mechanism changed: under Annex VIII Section B (consolidated text expected before 2 August 2026), providers self-classifying non-high-risk must register the system in the EU database with documented justification.
Modulos verbatim 7-8 May 2026: "the registration obligation survives in a streamlined form under Annex VIII Section B · converts self-assessment into a public artefact requiring documented justification". Lewis Silkin verbatim 7 May 2026: "providers must register AI systems in the EU database even where they consider their system exempt from high-risk classification".
Operationally: the self-classification is no longer a quiet internal choice. It becomes a public artefact — a deposited document, visible to regulators, peers, journalists, and counterparties. The quality of the documentation determines the defensibility of the position. This is exactly the type of artefact Sprinkling Act produces.
Sources: Reg. (EU) 2024/1689 Art. 6(3) + Annex VIII · Council/Parliament press release (7 May 2026) · Modulos.ai (7-8 May 2026) · Lewis Silkin (7 May 2026).
Four conditions
The four conditions of Article 6(3), verbatim
Article 6(3) lists four alternative conditions. One is sufficient — but the conditions are narrow, technical, and the defensibility of the self-classification depends on the documentation supporting the chosen condition.
(a) Narrow procedural task
The AI system is intended to perform a narrow procedural task. Examples in Commission guidance: classification (categorising incoming documents), structuring (rearranging unstructured data into a schema), transcription (speech-to-text). The narrower the task, the stronger the position. Generic systems performing broad tasks rarely qualify.
(b) Improvement of prior human activity
The AI system is intended to improve the result of a previously completed human activity. The human decision remains the operative one; the AI refines, edits, or augments. Common patterns: spelling/grammar checks on a human-drafted document, formatting suggestions, completeness checks. The AI does not initiate the decision.
(c) Pattern detection without replacing human review
The AI system detects decision-making patterns or deviations from prior decision-making patterns and is not meant to replace or influence the previously completed human assessment, without proper human review. This condition is procedurally demanding: pattern detection alone is insufficient — the absence of replacement/influence on prior human assessment must be documented, and proper human review must remain in place.
(d) Preparatory task to an Annex III assessment
The AI system performs a preparatory task to an assessment relevant for the use cases listed in Annex III. The system feeds into a separate, downstream assessment performed by a human or by another non-Annex-III system. The boundary between "preparatory" and "determinative" is where the defensibility test happens.
Sources: Reg. (EU) 2024/1689 Article 6(3) subparagraphs (a), (b), (c), (d) verbatim · Commission draft guidelines on Article 6 (forthcoming consolidated text post-Omnibus).
Our scope
What our assessment covers for Article 6(3)
The Sprinkling Act report for a system tested under Article 6(3) covers the six gates of the standardised methodology, with the 6(3) specificity layered in. The output: a documented methodological position, depositable as Annex VIII Section B justification, subject to legal validation by qualified counsel.
- →G1 — Art. 5 — Are any of the 8 prohibited practices triggered? (independent of 6(3) eligibility)
- →G2 — Art. 6(1) — Is the AI system a safety component of an EU-regulated product? (6(3) does not apply to 6(1) systems)
- →G3 — Art. 6(2) + Annex III — Does an Annex III use case apply? (entry condition for 6(3) test)
- →G4 — Art. 50 — Does an end user interact with the AI without knowing? (transparency obligation independent of 6(3))
- →G5 — Art. 51/53 — GPAI obligations (independent of 6(3))
- →G6 — Art. 6(3) — Mapping to the four conditions + profiling carve-out + indicators of potential eligibility
Article 6(3) specificity:
- ·Mapping of the system to conditions (a), (b), (c), or (d) — verbatim
- ·Profiling carve-out check (last subparagraph Article 6(3))
- ·Indicators of potential eligibility — never "eligible"
- ·Documented justification structured for Annex VIII Section B deposit
- ·Antecedence trail: timestamped record of the methodological position
- ·Boundary mapping: where preparatory ends and determinative begins (condition (d))
Out of scope (explicit):
- ×Final legal decision on non-high-risk self-classification (qualified counsel required)
- ×Filing of the registration in the EU database (operator's own act, post-counsel validation)
- ×Post-registration litigation, enforcement defence, or notified body interaction (separate engagement)
- ×Confirmation of "eligibility" — Sprinkling Act produces indicators, not determinations
Integration
How this fits with counsel and the compliance ecosystem
The Sprinkling Act Article 6(3) assessment produces the methodological documented justification. It does not replace: (i) the legal opinion of your qualified counsel on the final self-classification decision; (ii) the validation of your Notified Body where applicable; (iii) the registration filing that you operate yourself in the EU database under Annex VIII Section B.
Concretely: Sprinkling Act produces the public artefact, you file it, your counsel validates the legal defensibility. The article-by-article methodological position becomes the input to the legal review — not a substitute for it. The mention "Methodological assessment — not legal advice" is structural, not cosmetic.
Sources: Reg. (EU) 2024/1689 Art. 6(3) + Annex VIII Section B · Sprinkling Act Methodology v1.3 (May 2026) · structural distinction methodological/legal verified across all SA deliverables.
Test your position
The 9-question diagnostic identifies whether Article 6(3) is a potentially eligible path for your system — and surfaces the conditions and profiling carve-out that determine defensibility. 60 seconds. Zero data collected.
This page is informational. It does not constitute legal advice, regulatory determination, or a conformity assessment under Article 43 AIA. The Sprinkling Act assessment produces methodological indicators of potential eligibility under Article 6(3). The final self-classification decision rests with the provider and requires validation by qualified legal counsel before any deposit in the EU database under Annex VIII Section B.
SEE ALSO
Annex A — The Deployer Multiplier
Where the AI Act cascade becomes visible across deployer-side use cases.
Free Diagnostic
9-question assessment — Article 6(3) condition path flagged.
Sprinkling Act Methodology
The 6-gate framework behind the assessment.
Pricing
Free assessment. €690 full report. No subscription.