EU AI Act — Position Assessment
We Screened 50.
96% Have No Position.
Where Do You Stand?
Your stakeholders ask: “Where are you on AI Act?”
Full report guarantee: report doesn’t cover the system described in your intake, full refund.
Article 50 transparency · 2 August 2026 (new systems) then 2 December 2026 transitional (systems already on EU market)
Annex III high-risk · 2 December 2027 (Digital Omnibus trilogue agreement, 7 May 2026)
18-month runway for most deployers · 7-month window for GenAI providers with systems already on market
9 questions · 60 seconds · Your position across 6 regulatory gates.
No signup. No call. No waiting. Just your answer.
An independent document is different from a technical explanation.
Pre-conformity layer · Brussels-based (BCE BE 1034.962.482) · Independent · Client-funded · Research-indexed on Zenodo (3 publications, CC-BY 4.0)
FEATURED RESEARCH
Annex A — The Deployer Multiplier
May 2026 · 22 pages · 3 segments · cryptographically timestamped · DOI 10.5281/zenodo.20042174
Read the annex →BY THE NUMBERS
What the work measures
BY SEGMENT
Where the cascade hits first
Deployer segments where the AI Act post-Omnibus binding dates concentrate exposure. The 18-month runway is sector-specific — except for GenAI providers with systems already on the EU market (7-month window).
BY SEGMENT — 18-MONTH RUNWAY
BANKING / FINANCIAL SERVICES
Annex III §5 + DORA × Article 26
Annex III §5(b) · DORA · Binding 2 December 2027
Credit institutions using third-party AI scoring trigger Annex III §5(b) automatic FRIA. The 7 May 2026 Omnibus did not exclude banking. DORA × AI Act intersection sharpens with the Eurogroup demanding GPAI access for EU banks (Bloomberg, May 2026).
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. Postponed to 2 December 2027, 18 months to produce a defensible 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. Safety component definition narrowed, but medtech was excluded from the machinery overlap removal. MDR × AIA double-compliance persists.
GENAI PROVIDERS — 7-MONTH WINDOW
DEADLINE
2 December 2026 — transitional period
Article 50(2) watermarking obligations bind systems deployed before 2 August 2026 on 2 December 2026 (post-Omnibus transitional). Article 5 CSAM/NCII new prohibition: same date (Digital Omnibus trilogue agreement, 7 May 2026).
WATERMARKING
Multi-layered marking required
Code of Practice 1st draft: single-method approach is non-compliant. Required stack: metadata embedding (C2PA-style) + imperceptible watermark (SynthID-style) + post-edit detection. Final text: May–June 2026 Closing Plenary.
ENFORCEMENT
Civil liability already active
Dutch court ruling against Grok, 26 March 2026 (ECLI:NL:RBAMS:2026:3106): €100,000/day per defendant entity for non-consensual intimate imagery. Russmedia C-492/23 paradigm: platform operators bear primary responsibility for generative outputs. Civil and DSA liability is in motion before AI Act Art. 5 formally binds.
THE SITUATION
Where you stand, and what you're up against.
We built the opposite.
96% of the 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
15-20 pages, article-by-article mapping, reviewed by a human analyst. Delivered within 10 business days.
Dated · Signed · Verifiable badge
ENFORCEMENT TIMELINE
Feb 2025
Art. 5 applicable
Aug 2025
GPAI in force
Aug 2026
Art. 50 transparency
YOU ARE HERE
2027+
Full enforcement
THE ANSWER
The question has been asked.
Your investor asks. Your procurement team asks. Your regulator will ask. Not having a documented position does not mean having no position. It means having one you cannot defend. A dated, article-mapped, independently assessed proof, that’s what you send.
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. Biometric models. Real-time surveillance. These are not grey areas. They are the sectors where Art. 5 prohibitions are in force since February 2, 2025. Market surveillance authorities across all 27 EU member states are designated and empowered to enforce these obligations. Penalties: up to €35M or 7% of global turnover.
ISO 42001 is a voluntary management standard. EU AI Act 2024/1689 is a legal obligation. They address different questions. ISO 42001 certification does not establish your classification position under the AI Act. Sprinkling Act does.
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
12-month validity
Auto-renewal reminder at month 10
Embed ready
SVG badge + HTML snippet included with every report
The date on your report is permanent. It cannot be backdated, replicated, or faked. A document produced before enforcement is evidence of proactive positioning that stays in the regulatory record permanently. Like a timestamp in stone: the cost is one-time, the proof lasts as long as your organization exists.
One Document. One Price.
Two artifacts, two roles. The Free Diagnostic is the Nutri-Score of AI Act exposure: 60-second snapshot across 6 gates, instant readability, free. The Full Report is the EPC of AI Act: a dated, article-mapped boundary document delivered by an independent third party, opposable, defensible, anchored. The date cannot be backdated.
FREE
Free Diagnostic
€0
- ·6-gate boundary assessment
- ·Visual risk rating (star-based)
- ·Number of obligations identified
- ·Permanent shareable link
Score details, obligation list & legal references reserved for Full Report
RECOMMENDED
Full Report
€690
One-time · Pre-enforcement rate (€990 from 2 December 2027)
- ✓Everything in Free +
- ✓15–22 page 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
RECURRING
Signal Watch
€199/mo
€199/mo now → €299/mo from 2 December 2027
- ·Requires delivered Full Report
- ·Automatic re-scoring on regulatory changes
- ·Real-time classification shift alerts
- ·Monthly intelligence brief
INTELLIGENCE
Intelligence Report
€490/mo
€490/mo now → €690/mo from 2 December 2027
- ·Everything in Signal Watch +
- ·Sector-wide compliance benchmarks
- ·Competitor positioning analysis
- ·Quarterly regulatory deep-dive
- ·Full Sprinkling Act+ intelligence feed
Questions about scope or fit? Get in touch
CLARITY
What we are not.
We are not a law firm. Not a SaaS compliance platform. Not affiliated with any vendor, regulator, or certification body.
We do not produce legal opinions. We do not promise compliance.
What we are: an independent classification service that documents your regulatory position. 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
We don’t start from the assumption that your AI is a problem. Every AI deployed in an organization starts with a presumption of positivity, because rules exist, constraints apply, and within that framework it is not presumed guilty.
Descriptions, Not Systems
We don’t classify systems. We classify descriptions against the regulation as written. Every conclusion cites a specific article, not approximated, not interpreted. Position + proof = defensible governance.
The Boundary Layer
Sprinkling Act does not tell you what to do. It defines what your system cannot be without triggering regulatory obligations. Once the boundary is named, it cannot be un-named. That is the point.
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 →
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.
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?+
Sprinkling Act is an independent assessment service. It does not constitute legal advice and is not affiliated with or endorsed by the European Commission or any national supervisory authority. Assessment results are informational only.