Artificial intelligence and transparency

Legal basis

Regulation (EU) 2024/1689 of the European Parliament and of the Council (the AI Act) imposes transparency obligations in Article 50. They have applied since 2 August 2026.

Those obligations were not postponed by the 2026 amendments known as the digital omnibus - the postponement covered deadlines for high-risk systems, not the Article 50 transparency duties.

New European Strategies is not a provider of AI systems. We act as a deployer of tools supplied by third parties - and it is our obligations as a deployer that this document describes.

We operate no high-risk systems within the meaning of Annex III to the AI Act and no practice prohibited under Article 5 of the AI Act.

Where we use artificial intelligence

• Editorial support: suggested headlines, summaries, teasers and metadata descriptions, language editing, and draft translation between the Polish and English versions.
• Search and recommendations: selecting related content, ordering lists and matching topics to your interests.
• Assisted moderation: automated detection of spam and obvious abuse, holding content for human review.
• Internal tooling: summaries of source material for editorial and analytical work.
• Visual material: machine-generated illustrations used as conceptual artwork - never as documentary material.

How we label content

There is one principle: the reader should know what they are looking at before deciding how much to trust it.

• Text that is machine-generated and published WITHOUT human editorial review is labelled expressly as AI-generated - that is the requirement of Article 50(4) of the AI Act for text published to inform the public on matters of public interest.
• Material that has been edited and verified by a person, and for which we take editorial responsibility, does not require that label under the same provision - and that is how the overwhelming majority of our publications are produced. This does not mean no tools were involved: it means the editorial team is answerable for the result.
• Images, audio or video that are machine-generated or materially modified and depict people, events or places in a way that could appear authentic are labelled as artificially generated (Article 50(4) of the AI Act).
• Machine-generated conceptual illustrations are described in the caption. We never present them as documentary photography or as evidence of an event.
• Machine translations made available without human verification are labelled next to the content.
• We do not create or publish material depicting real people saying words they did not say, other than clearly labelled satire or educational material about disinformation.

What we never hand over to a machine

• Establishing facts. A factual claim in a publication is verified by a person against the source - a tool may propose it, it may not confirm it.
• Moderation decisions to remove content or restrict or suspend an account. A filter may hold content for review; a human takes the decision (/moderacja-komentarzy).
• Decisions on admission to a discussion club, on complaints and on GDPR requests.
• Emotion recognition and biometric categorisation - we use no such systems in any process.
• Assessing people's credibility or any scoring of participants in debate.
• Publishing under an author's name a text that the author did not edit and approve.

Conversational assistants

If we make an AI-based conversational assistant available, we will say so plainly at the start of the interaction - as Article 50(1) of the AI Act requires. We never imply that a human is on the other side.

Conversations between users in the service are conversations between people. We do not substitute a system into them and do not generate contributions on anyone's behalf.

Your data and AI tools

• We do not pass your personal data to model providers for training their models. We use these services on terms that exclude the use of submitted content for training.
• Private content - messages, club documents, reports and correspondence - is not sent to generative tools.
• Tool providers that process any data on our behalf are processors bound by a data processing agreement (Article 28 GDPR) and fall within the categories described in the data processing policy (/polityka-przetwarzania-danych).
• You can switch off recommendation personalisation by withdrawing the personalisation consent in privacy settings.

Errors, corrections and reports

Generative tools get things wrong in ways that sound convincing - which is exactly why human verification is not optional here.

• Report an error in a publication to office@neweuropeanstrategies.com. We publish corrections next to the material, with a date and a description of the change - we do not edit silently.
• If you believe material has been labelled incorrectly (as generated when it is not, or the other way round), write to us; we will fix the label or explain the basis for it.
• If your image or voice has been used in machine-generated material without your consent, we treat the report as a priority and take the material down while we investigate.
• Supervision of the AI Act rests with designated national authorities; independently of contacting us, you may approach the authority competent in your Member State.

Contact

Questions about the use of artificial intelligence in a specific piece or process: office@neweuropeanstrategies.com. We answer concretely - if a tool was used in a given publication, we will say what for.