AI Transparency Policy

#AI transparency policy of WPPoland.com

This policy explains how WPPoland uses artificial intelligence in producing content published on wppoland.com, who holds editorial responsibility for it, and how I label material that is generated or assisted by AI. I prepared the document with the transparency obligations of Article 50 of the EU AI Act, that is Regulation (EU) 2024/1689, in mind.

#§1. How I use AI in content production

1. I treat AI tools as support for editorial craft, not as an autonomous author. They help us with research, drafting article structure, and preparing visual material.
2. AI does not make a standalone decision to publish. Before anything goes live, a person reads it, corrects it, and approves it.
3. I do not publish content I cannot verify on the merits. Technical claims, figures, and legal references are checked against primary sources.

#§2. Editorial oversight and responsibility

1. All content published on the site undergoes human editorial review. The review covers factual accuracy, alignment with my practical experience, and consistency with my brand standards.
2. Editorial responsibility for published material rests with WPPoland Mariusz Szatkowski, as the publisher of wppoland.com.
3. Article 50(4) of the AI Act provides that the obligation to disclose AI use when generating or manipulating text published to inform the public on matters of public interest does not apply where the content has undergone human review or editorial control and a natural or legal person holds editorial responsibility. My process, grounded in human oversight and clear publisher responsibility, meets that condition. I publish this policy anyway, because I treat openness as a standard I hold ourselves to.

#§3. Labelling visual material

1. Some graphics on the site, including some author photos in post headers, were created entirely or substantially with AI. The author photos were generated from his real photos, with his consent. I do not put a caption under these images; this policy is where I disclose them.
2. Where technically feasible, I attach a machine-readable marking, in particular C2PA Content Credentials or appropriate file metadata, so that the origin of the material can be verified automatically.
3. Minor, assistive edits (cropping, exposure correction, noise reduction) that do not substantially change the meaning or context of the material are treated as ordinary processing and do not require separate marking within the meaning of Article 50.

#§4. No AI chatbot

1. The wppoland.com site does not operate a chatbot or any other AI system that interacts directly with users.
2. For that reason, the Article 50(1) obligation to inform people that they are interacting with an AI system does not apply to my site.
3. Should I add such a feature in the future, I will clearly inform users that they are talking to an AI system and update this policy.

#§5. The two-layer approach

For labelling AI content I apply a two-layer approach, in line with the structure of Article 50:

1. Machine-readable layer: C2PA Content Credentials, watermarking, or metadata, allowing machines to detect that material was generated or manipulated by AI, as far as is technically feasible.
2. Human-facing layer: a disclosure for the reader, which on this site is this policy rather than a caption under each image.

#§5a. Data and approved tools

1. AI tools may receive only the information needed for the approved task. Passwords, API keys, payment data and unpublished client material must not be entered into public AI services.
2. Client or user data may be processed with an AI service only when the service, account, contractual terms and retention settings have been approved for that purpose.
3. Where possible, data is removed, aggregated or replaced with test values before it reaches a model. Access follows the same least-privilege rule as other production systems.
4. AI output is not treated as a source. Factual statements must be checked against the referenced document, system or primary source before publication or use in a client deliverable.
5. Generated material remains subject to the normal deletion, access-control and project-retention rules. Moving work into an AI interface does not create a separate exception.

#§5b. Prohibited uses and incident handling

I do not use AI to fabricate testimonials, clients, statistics, credentials or project results; make autonomous decisions affecting a person’s legal position; publish content without human review; impersonate a named person; or bypass security and approval controls.

If an AI tool exposes protected information, produces harmful public content or acts outside its approved scope, I stop the affected workflow, preserve the relevant evidence, remove exposed access and identify the systems and records involved. Corrections, contractual notifications and any legally required reporting are handled according to the affected project’s incident procedure. The workflow returns only after the cause and prevention steps are recorded.

#§6. Legal basis and effective date

1. This policy addresses the transparency obligations of Article 50 of Regulation (EU) 2024/1689 (the AI Act).
2. The AI Act entered into force on 1 August 2024, and the Article 50 transparency obligations have applied since 2 August 2026.
3. Breaches of the Article 50 obligations fall under Article 99 of the AI Act and can reach up to EUR 15,000,000 or 3% of total worldwide annual turnover, whichever is higher.
4. The European Commission and the AI Office are finalising a code of practice and guidelines on the marking and detection of AI content. Once published, I will adapt my technical solutions to the detailed requirements.
5. In Poland, the process of designating a national supervisory authority for the AI Act is under way. Once its name is finalised, I will add it to this policy.

#§7. Related guides

I cover the labelling of AI content under Article 50 of the AI Act in detail in my guide: labelling AI-generated content under the AI Act.

This policy is part of a consolidated approach to EU compliance on WordPress projects; for a broader view across the AI Act, NIS2, DORA, and GDPR, see my EU compliance audit for WordPress.

#§8. Sources

1. Regulation (EU) 2024/1689 (the AI Act), text on EUR-Lex: https://eur-lex.europa.eu/eli/reg/2024/1689/oj.
2. European Commission, regulatory framework for AI: https://digital-strategy.ec.europa.eu/en/policies/regulatory-framework-ai.
3. C2PA (Coalition for Content Provenance and Authenticity): https://c2pa.org/.

#§9. Policy updates

1. I update this policy in response to legal, technological, or organizational changes in how AI is used on the site.
2. The current version is always available on wppoland.com.
3. Material changes are communicated on the Service.

#Contact

WPPoland Mariusz Szatkowski ul. Starowiejska 16/2 81-356 Gdynia, Poland e-mail: [email protected]

Related delivery work sits under WordPress developer.

#§10. Client work versus this site

This policy covers material published on wppoland.com. Client projects follow the contract and, where personal data is involved, an art. 28 GDPR processor agreement. By default I do not send trade secrets or personal data into public AI services.

If a client asks for AI-assisted drafting (for example page outline or alt-text proposals), a human still reviews the result and factual claims must be checked outside the model. Passwords, API keys, database dumps and full backups never go into a model.

#§11. How to ask about labelling

Questions about whether a specific asset on wppoland.com involved AI can go to [email protected]. The reply will describe AI’s role (if any), the editorial review, and C2PA status when available. There is no chatbot for these requests.