AI Guidelines, as of 17 June 2026

Our guidelines for the responsible use of AI

At a time of unprecedented technological disruption, media organisations must act proactively and establish clear rules for the responsible use of technology. Particularly in journalism, where integrity, objectivity and trustworthiness are of the utmost importance, strict guidelines must be followed.

Artificial intelligence and intelligent automation systems are key technologies that will fundamentally transform the way we create, distribute and consume content. Rather than fearing this change, we must actively shape and harness it. Alongside the highly dynamic development of these key technologies – though not always in step with it – legislation at EU and national level prescribes mandatory standards that must also be observed by companies using AI systems in their day-to-day operations.

For this reason, we have developed a dynamic set of guidelines that reflects our philosophy and approach to these transformative technologies. These guidelines should be understood as a living document undergoing constant further development, as we must remain flexible and adaptable in an environment where technical and regulatory requirements are changing rapidly.

Our overarching aim is to fulfil our mission as a trusted source of information and a driving force for society to the best of our ability. Artificial intelligence and automation are tools designed to support us in this by enabling efficiency gains and opening up new creative possibilities. Our guidelines address these challenges and set out clear parameters for the responsible use of these technologies. At the same time, they are aligned with the requirements of Regulation (EU) 2024/1689 (AI Act), in particular the transparency and labelling obligations (Article 50 of the EU AI Act). They reflect our determination to be pioneers in this transformation and to preserve our ability to act and make decisions.

Our guidelines for the use of AI include:

Safeguarding journalistic independence

Our core principle is, and remains, unbiased, fact-based reporting that provides guidance to our readers. AI systems are tools that support this objective, but must never undermine trust in our journalistic integrity.

Verification of information sources

All information sources and content undergo a rigorous verification process. Data that cannot be verified has no place in our reporting. AI-generated factual claims – particularly those from large language models – are not treated as sources, but are cross-checked against reliable primary sources in the same way as an unconfirmed hypothesis.

Editorial control and responsibility

Automatically generated content is subject to careful final review by qualified editors. Journalistic responsibility always lies with our editorial teams. Editorial responsibility for publication of the content rests with the competent editor-in-chief (Article 50(4), sub-paragraph 4, sentence 2 of the AI Act).

Open communication

We communicate transparently and openly with our readers about when and how AI technologies are used.

Compliance with ethical standards

The use of AI must not compromise our duty of care. The protection of sources and personal rights, as well as the handling of confidential input data (see the section below ‘Protection of confidential input data’), is always ensured.

Promoting creative freedom

The use of AI is intended to give our editors more scope for creative and journalistic activities. AI is therefore used to expand sources of information, research topics more quickly, develop ideas and, in practical application, simply to work faster and better.

Labelling of AI-generated content

Image, audio or video content that has been generated or manipulated using AI and resemble real people, objects, places or events are clearly labelled as such (Art. 50(4), first subparagraph, and (5) of the AI Act). The labelling appears directly on the content and is also visible when the content is shared via thumbnails and on social media. Transparency is key to ensuring that our readers can correctly understand the origin and nature of visual content.

The same applies mutatis mutandis to AI-generated or AI-manipulated texts on matters of public interest (Article 50(4), second subparagraph, of the AI Act). Provided that such texts have been subject to human review and editorial control, and editorial responsibility is assumed in accordance with the section ‘Editorial Control and Responsibility’, we invoke the exception under Article 50(4), fourth subparagraph, second sentence of the AI Act.

Protection of confidential input data

No content that could identify informants, disclose non-public sources or breach contractual confidentiality obligations shall be entered into third-party AI systems. The protection of informants and editorial professional secrecy (Section 53(1)(5) of the Code of Criminal Procedure [StPO], Section 383(1)(5) of the Code of Civil Procedure [ZPO]) apply without restriction.

AI Competence

Staff who use AI tools in an editorial capacity shall receive appropriate training on how the tools work, typical error patterns (in particular hallucinations) and the content of these guidelines (Article 4 of the AI Act).

Adaptation

These guidelines will be adapted to reflect legal and technical developments, in particular the Code of Practice of the European AI Office and the case law of the relevant market surveillance authorities.

By strictly adhering to these guidelines, we ensure that the responsible use of AI in the journalistic environment is guaranteed and that the high standards of our industry are upheld.