New legal requirements in Poland - Eco, Green and AI

Artificial intelligence increasingly supports the work of editorial offices, social organizations, businesses, educational institutions, and individual creators. We also use this tool.

Most of the content and graphics published on our website are prepared with AI support. We do this primarily because we are a small organization with a limited budget and human resources.

Regular preparation of materials requires time dedicated to:

  • searching for information;
  • analyzing many sources;
  • organizing data;
  • creating the article structure;
  • editing the text;
  • simplifying specialized language;
  • proofreading;
  • preparing illustrations;
  • publishing the material.

Without technological support, preparing such a number of materials would be much more difficult, slower, and more expensive for a small team.

AI is not a way for us to replace human responsibility. It is a tool that allows us to perform part of the time-consuming work faster and dedicate more attention to checking and refining the final material.

What exactly does AI do for us?

Artificial intelligence supports us at several stages of publication preparation.

Gathering and organizing information

AI helps to search for issues requiring verification, organize scattered information, and indicate areas that should be described in more detail.

However, we do not treat the system's responses as an independent source of knowledge.

As available, we primarily use:

  • publications of public institutions;
  • legal acts;
  • reports of international organizations;
  • scientific studies;
  • statistical data;
  • technical documentation;
  • materials from recognized industry organizations.

We try to indicate the sources used in the preparation of the article at its end, so that the reader can independently review the source materials.

Preparation of a draft version of the text

AI helps to create the initial structure of the article, organize arguments, and transform specialized information into language more accessible to the audience.

However, the first version is not automatically a ready publication.

The material may require:

  • shortening;
  • expansion;
  • removal of repetitions;
  • checking numbers and dates;
  • clarification of limitations;
  • changing examples;
  • linguistic improvement;
  • adaptation to our website's profile.

Artificial intelligence can prepare a convincingly sounding answer even when some information is incomplete, outdated, or incorrect. Therefore, the mere fluency of the text is not proof of its credibility for us.

Editing and simplifying language

Many important pieces of information are published in the form of legal acts, long reports, technical documentation, or specialized analyses.

AI helps to translate them into more understandable language.

Our goal is not to create complex scientific studies. We gather information and present it in a form that is intended to be:

  • accessible;
  • interesting;
  • practical;
  • usable in everyday life;
  • understandable without specialized education.

Simplification should not, however, lead to a change in meaning. Therefore, we try to retain the most important reservations, exceptions, and limitations resulting from the sources.

How do we use AI to create graphics?

A significant part of the illustrations published on our website are also created using generative tools; only a part are our original photos.

Graphics are primarily intended to:

  • attract attention to the topic;
  • illustrate the described phenomenon;
  • facilitate understanding of the issue;
  • build a consistent appearance for publications;
  • replace costly photo sessions and the purchase of stock materials.

A generated graphic does not have to depict a real place, event, or specific people. It is most often an illustrative representation of the article's topic.

This is particularly important for images resembling photographs. A realistic appearance does not mean that the depicted situation actually happened.

Therefore, for graphics prepared by AI, we may use the following information:

„Illustrative graphic prepared using artificial intelligence. It does not depict a documented event.”

Not every graphic generated using AI is legally treated as a deepfake. However, a situation where an image, recording, or video resembles authentic material concerning an existing person, place, object, or event and can be perceived as real is of particular importance.

Regardless of the minimum scope of legal obligations, we want to clearly inform readers that published illustrations may be synthetic.

Do we automatically publish what AI prepares?

No.

Content is not published directly after generation.

Before publication, the material goes through several stages:

Etap Co robimy? Goal
Topic selection we define the scope and purpose of the publication maintaining editorial direction
Material collection we search for information and sources establishing substantive foundations
Draft version we use AI to organize content accelerating work
Verification we check key facts, data, and conclusions reducing the risk of errors
Editing we improve language, structure, and examples adapting the material to the audience
Approval a human makes the decision to publish maintaining editorial responsibility

The final material is therefore the result of human collaboration with a tool, not an autonomous system publication.

Who is responsible for the content?

Responsibility for topic selection, the manner of presenting the issue, and the decision to publish remains with us.

We do not transfer it to the AI tool.

This also applies to situations where the system:

  • provides outdated information;
  • incorrectly interprets a source;
  • combines incomparable data;
  • omits an important exception;
  • prepares too categorical a conclusion;
  • generates a graphic element that may be misleading.

Human control does not guarantee the complete elimination of all errors. However, it allows for reducing risk and responding to reported comments.

If an error is noticed, the reader can bring it to our attention. Justified corrections should be implemented as quickly as possible.

Why are we informing about this now?

As of August 2, 2026, the provisions of Article 50 of the EU AI Act concerning the transparency of certain systems and content generated or modified by AI are applicable.

Regulations include, i.e.:

  • informing a human when they enter into direct interaction with an AI system;
  • technical labeling of synthetic content by system providers;
  • disclosing the use of AI in the case of deepfakes;
  • labeling certain texts concerning matters of public interest.

However, this does not mean that every text prepared with the use of AI must automatically receive a separate label.

With regard to texts concerning matters of public interest, Article 50 provides for an exception when the material has undergone a human verification process or editorial control, and a specific natural or legal person bears editorial responsibility for the publication.

In our case, AI serves an auxiliary function, and materials are checked, corrected, and approved before publication.

Despite this, we decided to openly inform about our working method. We do this not only because of the regulations, but primarily because the reader should know how the materials provided to them are created.

What has the Polish law on artificial intelligence systems changed?

The Act of July 3, 2026, on artificial intelligence systems does not introduce a simple obligation to label every text and every graphic prepared with the help of AI.

Its main task is to create a national system for the application and supervision of the EU AI Act.

The Act specifies, i.e.:

  • the organization of supervision over the market of AI systems;
  • the competences of national authorities;
  • procedures in case of infringements;
  • rules for control;
  • the possibility of imposing administrative monetary penalties;
  • actions supporting the development and safe use of AI.

Basic principles of content transparency stem directly from the EU regulation, while the Polish law organizes the national system of supervision and enforcement of these regulations.

What does AI not replace for us?

Artificial intelligence does not replace:

  • editorial responsibility;
  • independent thinking;
  • expert knowledge;
  • source verification;
  • practical experience;
  • decisions requiring individual analysis;
  • legal, medical, financial, or technical consultation.

Materials published on the website are primarily informative, educational, and popularizing in nature.

They should not be treated as individual expertise adapted to all circumstances of a specific case.

In matters where an error can lead to serious consequences, it may be necessary to seek the help of a qualified specialist.

AI helps us do more with limited resources

For a large editorial office, preparing an article can mean the work of several people: an author, a researcher, an editor, a proofreader, a graphic designer, and a person responsible for publication. In a small organization, these roles are often performed by one or several people, simultaneously with many other duties.

AI partially helps to equalize this difference.

Thanks to this, we can:

  • address more important topics;
  • analyze available materials faster;
  • publish regularly;
  • prepare our own illustrations;
  • improve text readability;
  • reach more people with knowledge.

This does not mean that every application of AI is flawless or risk-free. It means that with human control, technology can be an extremely helpful tool for an organization that does not have an extensive team and a large budget.

Our principle for using AI

We can describe it most simply as follows: Most of the content and graphics published on our website are prepared with the support of artificial intelligence. AI helps us search and organize information, develop draft texts, improve their readability, and create illustrations. Every material is checked, corrected, and approved by a human before publication. Responsibility for the final content remains on our side.

We do not want to present AI either as a threat that should be completely rejected or as a solution that can be unconditionally trusted.

It is a tool.

Its value depends on the method of use, data quality, human control, and readiness to correct errors.

What is most important for us is that the final material is interesting, understandable, useful, and as well-founded as possible on reliable sources.

Don't just like, act!

Źródła

  1. European Parliament and Council of the European Union — Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence – full text of the AI Act, including Art. 50 concerning the transparency of content generated and modified by AI.
  2. European Commission — Guidelines on transparency obligations under Art. 50 of the AI Act – overview of the scope of obligations, exceptions, and the importance of human editorial control.
  3. European Commission — Quick Facts: Transparency rules for AI systems – summary of principles concerning text, images, recordings, deepfakes, and content subject to human control.
  4. European Commission — Code of Conduct on labelling content generated by AI – practical solutions supporting the implementation of transparency obligations binding from 2 August 2026.
  5. Journal of Laws — Act of 3 July 2026 on artificial intelligence systems, Dz.U. 2026 item 1003 – national rules for supervision over.
  6. Ministry of Digital Affairs — Implementation of the AI Act in Poland – information on the direct application of the EU regulation and national implementation measures.

The content and graphics were prepared with the support of artificial intelligence tools, and then verified, edited, and approved by a human; editorial responsibility for the publication rests with IGTSF.

Legal basis and reference: Article 50(4) of Regulation (EU) 2024/1689 of the European Parliament and of the Council on artificial intelligence, applicable from 2 August 2026, and the Act of 3 July 2026 on artificial intelligence systems, Journal of Laws 2026 item 1003. The annotation also constitutes a voluntary transparency standard — with actual editorial control, not every AI-supported text is subject to the obligation of individual marking.

DONATION
Support our mission

Every donation helps implement environmental protection projects. Join the ranks of supporters!

Support us now

New legal requirements in Poland - Eco, Green and AI

Planted? Share

Get in touch with us!

Have questions or want to find out more about the partnership? Fill in the form and we will contact you to address your needs and discuss partnership opportunities.

Contact us!

Do you have questions or want to learn more about cooperation? Fill out the form, and we will contact you to address your needs and discuss partnership opportunities.