Where do you stand on the AI Act?
96% of the companies we screened can’t show it in writing. Can you?
The documented proof — we produce it: Timestamped · Defensible · Mapped article by article.
Free Diagnostic · Under 60 Seconds · No Account Required
Issuing body
Sprinkling Act
BCE BE 1034.962.482 · Bruxelles
Date issued
12.06.2026
Réf. SA-2026-0481
Assessment result
High risk
Independent party · Art. 26 AI Act
Applicable obligations
Guarantee (Art. 9bis ToS): if the report doesn't cover the system described in your questionnaire, full refund within 14 days.
Pre-conformity layer · Brussels-based (BCE BE 1034.962.482) · Independent · Client-funded · Research-indexed on Zenodo (3 publications, CC-BY 4.0)
AI ACT SCOPE
You use AI to make decisions about people? You are likely in scope.
Annex III — the most tightly regulated category. Few deployers have documented it.
88%
88% of organizations use AI. Fewer than one in five have minimum governance practices in place.
McKinsey State of AI 2025 · 2025 AI Governance Survey (Pacific AI, n=351)
Our report settles your case, in writing: in scope or not, and on which articles. Run the diagnostic →
BY THE NUMBERS
What the work measures
THE OBVIOUS QUESTION
Why not just ask an AI?
CONCESSION
You should. An AI analyzes your system against the AI Act faster and cheaper than any consultant.
THE LIMIT
What it cannot do: sign the result. A self-assessment is worth nothing to procurement or a regulator — not because it's wrong, because it comes from you.
WHAT YOU BUY
What you buy: your AI Act position, dated and independent. What no AI output or internal memo can ever be.
BY SEGMENT
Where the cascade hits first
The sectors AI Act binding deadlines hit first. 18-month runway — except GenAI providers already on EU market (7 months).
- Art. 5In force since Feb 2025
- GPAIIn force since Aug 2025
- Art. 50Effective Aug 2, 2026
Art. 50 transparency · 2 Aug 2026 (new) · 2 Dec 2026 (transitional)
- Annex IIIEffective 2 Dec 2027 (high-risk)
Annex III high-risk · 2 December 2027 (Digital Omnibus)
18-month runway for deployers · 7-month window for GenAI providers on EU market
BY SEGMENT · 18-MONTH RUNWAY
BANKING / FINANCIAL SERVICES
Annex III §5 + DORA × Article 26
Annex III §5(b) · DORA · Binding 2 December 2027
Third-party AI scoring triggers Annex III §5(b) automatic FRIA. Omnibus 2026 did not exclude banking. DORA × AI Act tighten simultaneously.
HRTECH / EMPLOYMENT
Annex III §4 · Employment AI
Annex III §4 · Binding 2 December 2027
AI used in hiring, screening, performance, or termination falls under Annex III §4. 18 months to document your position before binding obligations apply.
HEALTHCARE / MEDDEV
Article 6(1) MDR/IVDR pathway
Art. 6(1) · MDR/IVDR · Binding 2 August 2028
Medical AI under MDR/IVDR: postponed to 2 August 2028. Dual compliance MDR/IVDR + AI Act maintained; machinery overlap removal did not apply to medtech.
GENAI PROVIDERS · 7-MONTH WINDOW
DEADLINE
2 December 2026 · transitional period
Art. 50(2) watermarking + Art. 5 new prohibitions (CSAM/NCII): same date. Post-Omnibus transitional (trilogue agreement, 7 May 2026).
WATERMARKING
Multi-layered marking required
Single-method approach is non-compliant. Required stack: C2PA metadata + SynthID imperceptible watermark + post-edit detection.
ENFORCEMENT
Civil liability already active
Dutch court vs. Grok: €100,000/day (ECLI:NL:RBAMS:2026:3106). Russmedia C-492/23: primary liability on generative outputs. Civil + DSA already in motion before Art. 5.
THE SITUATION
Where you stand, and what you're up against.
We built the opposite.
Most companies we screened had no documented answer. The compliance industry made it opaque, slow, dependent, and vague.
Blind
You deploy AI. You’ve never heard of the AI Act. No one can name a single article.
Reassured
“We just use an API.” “Our lawyer handles it.” You assume it doesn’t apply. It does.
Informed
You know the AI Act exists. But you haven’t classified a single feature. No document. No date. No proof.
Documented
Every AI feature classified: Art. 5, 6, 50, GPAI. A dated, article-mapped artifact. Your investor asks. You answer in 30 seconds.
You pay a firm. You receive a PDF. You don’t understand why you were classified that way.
Every conclusion cites a specific article. Every gate shows its reasoning. You see exactly why.
Weeks of back-and-forth before you get a first answer. By then the deadline moved closer.
9 questions. Instant result. Full report within 10 business days. No discovery call. No waiting list.
Every question costs a call. Every update costs a bill. You never own the answer.
Your report is yours permanently. Dated. Signed. Verifiable. You don’t need us to read it, use it, or share it. Updates are optional, never mandatory.
Recommendations but no proof. “You should consider…” is not an artifact your investor can verify.
Article-mapped. Dated. Badge-verified. A structured input for your lawyer’s analysis, your auditor’s review, and your board’s decision.
The same actor sells the tooling and tells you what passes. Your investor cannot separate the score from the supply.
Sprinkling Act sells nothing else. No tooling, no SaaS integration, no implementation services. The classification stands on its own grid, separable from any builder, defensible to a qualified third party.
We screened 50 European AI companies (April 2026). 96% have no documented AI Act position. They’re operating at Level 1–3. Only 2 mention it publicly.
HOW IT WORKS
From diagnostic to documented position.
Three steps. No ambiguity. Your position is determined by the regulation, not by our opinion.
Free diagnostic
9 questions, 60 seconds. Covers 6 regulatory gates: prohibited practices, high-risk, GPAI, transparency.
Instant result, no account required
Detailed intake
A structured questionnaire appears in your dashboard. It adapts to your declared role and system. Your answers drive the classification.
Accuracy is required
Your report
Article-by-article mapping reviewed by a human analyst, adapted to your classification. Delivered within 10 business days.
Dated · Signed · Verifiable badge
THE 6 REGULATORY GATES
Six Gates. One Boundary.
High Risk
Example score
Like a Nutri-Score for AI regulation. 9 questions, 60 seconds. You receive a risk grade, a score out of 100, and a permanent shareable link.
Prohibited Practices
Art. 5
8 practices checked: if triggered, immediate cessation
If triggered: your product cannot be placed on the EU market. Penalties up to €35M or 7% of global turnover.
Safety Component
Art. 6 §1
Critical component in an EU regulated product. MDR Class IIa+ = automatically high-risk under AI Act with no further analysis needed (MDCG 2025-6).
If triggered: Art. 11-15 obligations (technical documentation, logging, human oversight). May delay your CE marking process.
High-Risk Sector
Art. 6 §2
8 Annex III sectors + Art. 6(3) exception test
If triggered: mandatory conformity assessment before market placement. May impact your product roadmap and fundraising timeline.
GPAI Systemic Risk
Art. 51
10²⁵ FLOPs threshold: adversarial testing required
If triggered: adversarial evaluations, AI Office reporting, enhanced cybersecurity obligations. Relevant during investor due diligence.
Transparency
Art. 50
Human interaction, deepfakes, content synthesis
If triggered: your users must know they are interacting with AI. Non-compliance may result in user complaints and media exposure.
GPAI Standard
Art. 53
Standard obligations for all general-purpose AI models
If triggered: technical documentation, copyright policy, training data summary required. Your B2B clients may start requiring this contractually.
HR algorithms, emotion detection, biometrics, real-time surveillance: Art. 5 in force since February 2, 2025. Penalties: up to €35M or 7% of global turnover.
ISO 42001 is a voluntary management standard. It does not determine your classification under the AI Act — Sprinkling Act does.
Report Preview
Adapted to your classification · Delivered within 1–2 weeks
SAMPLE REPORT
Screening Report
EU AI Act
Position Assessment Report
HIGHSprinkling Act · BCE 1034.962.482
EU AI Act Compliance · Brussels
| Client | Acme Intelligence BV |
| Country | EU · Netherlands |
| Role | Provider |
| Date | March 2026 |
EXECUTIVE SUMMARY · DETACHABLE
Acme Intelligence BV
Provider · EU · Netherlands
Classification
HIGH
Max. sanction
15M€ / 3%
Deadline
08/2027
The AI system has been assessed through the 6 regulatory gates of the Sprinkling Act framework. Classification HIGH based on Annex III sector detection with full Art. 9-15 obligations applicable.
+12 pages of detailed analysis
Obligations, remediation, GDPR Art. 22, bias, AI Positive, timeline, FAQ
Get Started · Free DiagnosticReal anonymized report. Adapted to your classification, delivered in 1–2 weeks. SVG visualizations, legal notices, integrated FAQ.
THE DIFFERENCE THAT MATTERS
Claims vs. documented proof.
Moving beyond unverifiable statements, into dated, article-mapped, independent evidence.
The Old Way
The New Way
A Sprinkling Act report is not a claim. It is a dated, article-mapped, independently assessed proof, the document you send when procurement, investors, or regulators ask.
THE BADGE
Proof your prospects can verify.
Self-assessment gets you started. An independent, auditable document gets you through procurement. Every Full Report includes a verifiable badge, dated, linked to your public assessment.
ASSESSED
SA
2026
Assessed by Sprinkling Act
AI Act Position Assessment · Assessed
Verifiable link
Anyone can check your score and assessment date
Risk-based validity
About 3 months (high-risk) to 12 months (moderate); sooner on any material change
Embed ready
SVG badge + HTML snippet included with every report
One Document. One Price.
FREE
Free Diagnostic
Nutri-Score of AI Act exposure: 60-second snapshot across 6 gates, instant readability.
€0
- ·6-gate boundary assessment
- ·Visual risk rating (star-based)
- ·Number of obligations identified
- ·Permanent shareable link
- ·Private: no account, no data collected
Score details, obligation list & legal references reserved for Full Report
RECOMMENDED
Full Report
EPC of AI Act: dated, article-mapped, by an independent third party. Defensible, and the date cannot be backdated.
€690
One-time · excl. VAT 21%
- ✓Everything in Free +
- ✓In-depth report, adapted to your classification
- ✓Detachable one-pager with risk gauge + visual timeline
- ✓GDPR Art. 22 cross-analysis + bias assessment
- ✓AI Positive governance radar + remediation plan
- ✓Official badge + priority support
RECURRING
Signal Watch
€249/mo
6-month commitment
€349/mo
No commitment
Subscribers keep their rate. Permanently.
The report captures your position once. Signal Watch keeps it current: re-scored at every regulatory shift, so your documented position never goes stale.
- ·Requires delivered Full Report
- ·Re-scoring at every regulatory shift
- ·Real-time classification shift alerts
- ·Monthly intelligence brief
- ·Each re-score delivered as a signed, timestamped attestation (open-source engine, Ed25519 + RFC 3161)
Questions about scope or fit? Get in touch
CLARITY
What we are not.
Not a law firm, not a SaaS platform, not affiliated with any regulator.
No legal opinions. No compliance promises.
An independent classification service. The starting point, not the destination.
THE FIRST STEP
The boundary comes first. Everything else follows.
The document that exists before the lawyer, the auditor, or the regulator asks.
Presumption of Positivity
Your AI isn’t presumed guilty.
The rules apply. Within that framework, your system is legitimate until classified — not guilty by default.
Descriptions, Not Systems
We classify what your AI does, not its name.
What your system does, in your own words, against the regulation as written. Every conclusion cites a specific article.
The Boundary Layer
We define what your system legally cannot be.
What your system cannot be without triggering obligations. Once documented and dated, it’s on the record.
The AI Positive framework is integrated into every Sprinkling Act Full Report. Learn more about the AI Positive framework →
Building AI agents? The EU AI Act already covers them. Read the 4 ACTS framework →
FAQs
We’re a US company. Does this apply to us?+
We’re ISO 42001 certified. Aren’t we covered?+
What about Art. 4 AI literacy?+
Your Position. Documented.
9 questions. 6 gates. One boundary. The date cannot be backdated.
Start Your AssessmentThe date on your report cannot be backdated.
Or explore the AI Act Compliance Index, the public registry of documented positions.
Regulatory signals, when they happen.
Unsubscribe anytime · no third-party sharing. Need monitoring alerts? Signal Watch.