Skip to main content

Built by a dual Canadian–American citizen. EEA data routed through Canada. Solo founder, replies may run slow.

EU AI Act transparency disclosure

Regulation (EU) 2024/1689 · Article 50 · Published July 26, 2026

Status: this is our own good-faith assessment, published for transparency. It is not legal advice, not a certification, and not a regulatory determination. Renew EcoMe LLC is a solo-founder company and this disclosure has not yet been reviewed by independent EU counsel. Where a question is genuinely open, it is marked pending counsel rather than answered.

SmarterTariff uses artificial intelligence to help produce its findings. We publish where AI is used, what its output means, and where it stops being reliable. We hold our own product to the standard we ask of the sites we assess.

1. What applies, and when

Article 50 sets transparency duties and applies from 2 August 2026. The Digital Omnibus on AI, adopted by the European Parliament on 16 June 2026 and the Council on 29 June 2026, postponed several high-risk obligations into 2027 and 2028. It did not postpone Article 50. We note this because the two are often reported together and the distinction matters.

Systems placed on the EU market before 2 August 2026 have a four-month grace period, to 2 December 2026, for the machine-readable marking duty in Article 50(2). SmarterTariff has been operating since March 2026 and we treat it as falling inside that window.

2. Our role

The AI Act assigns duties by role, and role is assessed per system rather than per company. For SmarterTariff:

  • We are the provider of the SmarterTariff assessment system, which we develop and offer under our own name.
  • We are a deployer of a third-party general-purpose AI model, which we call through an authenticated server-side API. We do not train, fine-tune, or host that model, and we are not its provider.

3. Risk tier

Assessed against the Act’s categories, in order:

  • Article 5 (prohibited practices): none apply. The system performs no social scoring, no biometric identification or categorisation, no emotion recognition, and no predictive policing.
  • Annex III (high-risk): in our assessment none apply. The system assesses the accessibility and agent-readiness of publicly reachable business web pages. It does not determine any natural person’s access to education, employment, credit, insurance, essential services, migration status, or justice. Pending counsel.
  • Article 50 (transparency): engaged. The system presents AI-assisted written findings to the person who requested them, so we treat it as carrying transparency duties.

Our working classification is therefore limited risk, with Article 50 transparency obligations. We would rather state that plainly and be corrected than claim a lighter classification than we can defend.

4. How we meet Article 50

  • 50(1), you are told it is AI. Findings are labelled as AI-assisted in the product and here. We do not present them as human review.
  • 50(2), marking generated content. Written findings are generated with AI assistance. We are implementing machine-readable marking within the grace period described above. Pending counsel on the precise form the marking must take for advisory text of this kind.
  • 50(3), emotion recognition and biometric categorisation. Not applicable. The system does neither.
  • 50(4), deep fakes and public-interest text. Not applicable. Output is advisory analysis returned to the requesting user, not synthetic media and not text published to inform the general public.
  • Article 4, AI literacy. Applicable since February 2025. As a solo-founder company, the person operating the system is the person who built it.

5. What the output is, and is not

  • Advisory, not certification. Findings are informational. They are not a certification, attestation, warranty, or guarantee of compliance with any law or standard, including the European Accessibility Act, WCAG 2.2 AA, EN 301 549, or the GDPR.
  • Not legal advice. Findings do not substitute for review by a qualified professional.
  • Human review is expected. AI output supports human judgement. Automated results should be supplemented with manual testing before decisions are taken.
  • We report what we can observe. The system reports the status of a public page. It does not offer an opinion on the operator of that page. Where it cannot see something, it says so rather than inferring a result.

6. Data and processing

AI processing runs server-side through authenticated calls. It does not run in your browser. Scanning reads publicly reachable pages and honours robot exclusion signals. API traffic from the EEA is routed through Canadian infrastructure, which the European Commission recognises as providing adequate protection under GDPR Article 45. Personal data handling is described in our Privacy Policy and Data Processing Agreement.

7. Contact

Questions about this disclosure, including from a supervisory authority, go to chris@smartertariff.com. Renew EcoMe LLC, Cape Elizabeth, Maine, USA. Sole member: Christopher M. Edwards.

See also our Canadian AI and automated-decision posture, our EAA statement, and our accessibility statement.