AI Transparency and Editorial Responsibility

Last updated: 2 August 2026

Our Approach to the Responsible Use of Artificial Intelligence

The Léoré Group stands for knowledge, dialogue and impact. Trust, substantive content, diligence and clearly attributable responsibility form the basis of our editorial and communications work.

Artificial intelligence can provide useful support for editorial, linguistic, creative and organisational processes. At Léoré, however, it does not replace human judgement or editorial responsibility.

Léoré uses ChatGPT Business by OpenAI for selected work processes. The application is used within a business workspace and as part of our internal editorial, confidentiality and data-protection procedures.

According to information published by OpenAI, inputs and outputs from ChatGPT Business are not used by default to train or improve OpenAI’s models. OpenAI also states that business data is encrypted both in transit and at rest and that contractual data-protection arrangements are available for business customers.

Irrespective of these technical and contractual safeguards, all content intended for publication is reviewed, edited and approved by a responsible person. Editorial and data-protection responsibility remains with Léoré at all times.

These Transparency Principles explain how Léoré uses AI-supported tools, which editorial safeguards apply and under which circumstances content may be labelled as AI-generated or AI-manipulated.

1. Scope

These principles apply to publicly accessible content produced by the Léoré Group and its business divisions.

This includes, in particular:

• articles and publications by Léoré Media;

• analyses, columns and editorial profiles;

• interviews and discussion formats;

• content published on websites and social media platforms;

• newsletters and digital publications;

• publication and communication formats developed by Léoré Studio;

• editorial partner formats;

• images, illustrations, graphics, audio recordings and videos.

AI-supported tools may also be used for internal working documents, unpublished drafts and organisational processes. Such use is governed by the applicable confidentiality, security and data-protection requirements.

These Transparency Principles describe the editorial and organisational use of artificial intelligence. They do not contain complete information on the processing of personal data. For such matters, the separate Privacy Policy of Léoré applies.

2. AI System Used

ChatGPT Business by OpenAI

As part of its editorial, linguistic and organisational work, Léoré uses ChatGPT Business, a version of ChatGPT provided by OpenAI for business use.

At Léoré, ChatGPT Business is used as a supporting work tool. It does not assume independent editorial control, make legally binding decisions or take responsibility for publications.

Outputs generated by the system are generally treated as working material. They are not automatically regarded as factually correct, complete, suitable for publication or editorially approved.

The use of ChatGPT Business does not in itself establish an official partnership, endorsement or institutional cooperation between Léoré and OpenAI. OpenAI is identified as the provider of the technical service used.

3. Data Protection when Using ChatGPT Business

3.1 Protection of Business Data

According to information published by OpenAI, the following settings and safeguards apply to ChatGPT Business:

• inputs and outputs from the business workspace are not used by default to train or improve OpenAI’s models;

• business data is encrypted in transit and at rest;

• members of a Business workspace generally have their own individual chat histories;

• other workspace members do not automatically have access to all conversations of another user;

• contractual data-protection arrangements are available for business customers.

These features distinguish the business workspace used by Léoré from an uncontrolled use of private or freely accessible AI accounts.

3.2 Use in Accordance with Data-Protection Requirements

Compliance with data-protection law does not arise solely from the name or technical design of a product. It also depends on how the system is used in practice.

Léoré therefore organises the use of ChatGPT Business in accordance with relevant data-protection principles, including:

• purpose limitation;

• data minimisation;

• confidentiality;

• access control;

• assessment of necessity;

• appropriate technical and organisational measures;

• examination of the applicable legal basis;

• documentation of relevant processing activities.

As a general rule, only information necessary for the respective editorial or organisational purpose is processed.

Personal, confidential or particularly sensitive information is not entered into the system without prior review. Where possible, information is reduced, anonymised or pseudonymised.

3.3 Data Used in Connection with Articles

In connection with editorial work, the following content may be processed using ChatGPT Business:

• publicly available information;

• text drafts created by Léoré;

• outlines and editorial concepts;

• wording proposals;

• working titles and headlines;

• approved background information;

• text passages requiring linguistic revision;

• information for which an appropriate legal basis for processing exists.

Particular protection applies to:

• unpublished interviews;

• confidential background discussions;

• private contact details;

• sensitive personal information;

• internal documents belonging to partners or customers;

• trade and business secrets;

• statements not intended for publication;

• application or employment-related data;

• access credentials and security information.

Such information may only be processed where this is necessary, legally permissible and compatible with Léoré’s internal confidentiality and data-protection requirements.

3.4 No General Guarantee of Data-Protection Compliance

Léoré does not describe ChatGPT Business as being automatically compliant with the General Data Protection Regulation in every conceivable use case.

Instead, the following principle applies:

Léoré uses ChatGPT Business within a controlled data-protection process and in accordance with the requirements of the General Data Protection Regulation.

Responsibility for selecting the information submitted, determining the legal basis, configuring the system, managing internal access rights and using the results editorially remains with Léoré.

For all questions concerning the processing of personal data, only the current version of Léoré’s Privacy Policy shall apply.

4. Areas in Which Artificial Intelligence May Be Used

4.1 Research Support

AI systems may assist in structuring subject areas, developing possible research questions, organising publicly available information or identifying starting points for further research.

Information generated by an AI system is not regarded as correct, complete or reliable merely because it has been produced by the system. Relevant facts, sources, names, figures, quotations and contextual relationships are reviewed by the editorial team.

4.2 Structuring and Concept Development

AI may be used to develop outlines, working titles, interview questions, thematic approaches, editorial perspectives or initial concepts.

The final selection, weighting and direction of an article are determined by a responsible person.

4.3 Creation of Working Drafts

ChatGPT Business may produce initial drafts, wording suggestions or alternative presentations.

At Léoré, such outputs are treated as editorial working material. Before publication, they are reviewed, contextualised, edited and expressly approved.

The direct and unreviewed publication of AI-generated output does not meet Léoré’s editorial standards.

4.4 Linguistic Editing

AI may be used to improve readability, grammar, spelling, style, clarity and linguistic consistency.

The meaning, editorial assessment and responsibility for the final content remain with Léoré.

4.5 Translation

AI-supported translation tools may be used for multilingual content.

Relevant translations are reviewed before publication for meaning, clarity, tone and possible cultural misunderstandings. Legally, professionally or socially sensitive content may be subject to additional human linguistic or specialist review.

4.6 Transcription and Summarisation

Conversations, interviews, meetings or audio recordings may be transcribed or summarised using technical systems for further editorial work.

Such processing only takes place in accordance with the necessary consent requirements, agreements, confidentiality obligations and data-protection rules.

4.7 Image, Audio and Video Processing

AI may assist with the technical processing of media content produced or controlled by Léoré, including:

• subtitling;

• transcription;

• image or audio optimisation;

• noise reduction;

• cropping;

• format conversion;

• colour or sound correction;

• accessibility improvements;

• removal of purely technical or distracting elements.

Where content is artificially generated or altered in a way that could create a false impression of real people, events, locations or statements, it will be labelled accordingly.

5. Human Editorial Control

Content published by Léoré is generally subject to human editorial review and approval.

This review is not limited to spelling, grammar or formatting. Depending on the nature, subject matter and significance of the content, the editorial process may include:

• reviewing the central statement;

• verifying relevant facts;

• assessing the sources used;

• checking names, functions, figures and dates;

• reviewing direct and indirect quotations;

• editorial contextualisation and weighting;

• revising structure and argumentation;

• adapting tone and language;

• checking for misleading or ambiguous statements;

• examining possible conflicts of interest;

• obtaining specialist or legal input where necessary;

• final approval by a responsible person.

The editorial team has the authority to amend, postpone or completely reject content.

A text is not published solely because it was generated or suggested by ChatGPT Business or another technical system.

Where AI is used to prepare a draft, the final content must be shaped by meaningful human editorial work. A purely superficial or formal review is not equivalent to substantive editorial control.

6. Editorial Responsibility

Responsibility for published content remains with Léoré or with the relevant editorial team, author or expressly designated publisher.

ChatGPT Business and other AI systems are not treated as authors in the journalistic or legal sense. An AI system cannot:

• assume editorial responsibility;

• independently guarantee the reliability of a source;

• legally approve a publication;

• accept liability for incorrect information;

• replace human interpretation and judgement.

The use of an AI-supported tool does not release Léoré from its obligation to review content carefully and assess publication responsibly.

Our editorial approach is guided by the following principles:

Substance before volume

Every publication must have a clear and legitimate editorial purpose.

Dialogue before premature judgement

Léoré asks questions, provides context and opens perspectives.

Impact before mere reach

Attention should support orientation, trust or social value.

Transparency before presentation

Cooperations, roles, interests and relevant technical interventions should remain understandable.

Human responsibility before automation

Technical support does not replace editorial decision-making.

7. Facts, Sources and Possible Errors

Generative AI systems may produce inaccurate, incomplete, outdated or invented information. They may oversimplify contexts, misattribute sources or generate convincing statements without a reliable factual basis.

Léoré therefore does not treat AI-generated output as independent evidence or as a reliable primary source.

Where an article contains verifiable factual statements, appropriate sources are used wherever possible. These may include:

• official documents;

• legislation and regulatory texts;

• original scientific sources;

• publications by public authorities;

• primary data and annual reports;

• direct statements by the persons involved;

• documented interviews;

• verifiable press releases;

• publications by recognised institutions.

The scope of the review depends on the subject matter, current relevance, public significance and potential consequences of an incorrect publication.

Despite careful editorial work, errors cannot be entirely excluded. Substantiated notifications of possible errors will be reviewed. Necessary corrections or additions will be made following editorial assessment.

8. Labelling of AI-Generated Content

Léoré labels content where labelling is required under applicable law or where it appears appropriate for editorial transparency.

The use of ChatGPT Business does not automatically mean that every supported article is labelled as AI-generated.

The relevant factors include:

• the nature and extent of the AI use;

• the degree of human editing;

• the existence of substantive editorial control;

• responsibility for the publication;

• the medium and form of presentation;

• the risk of deception or misunderstanding;

• applicable statutory labelling requirements.

Léoré takes into account the transparency requirements of Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence, commonly referred to as the EU Artificial Intelligence Act or EU AI Act.

In particular, Article 50 of the EU AI Act establishes transparency requirements for certain AI-generated or AI-manipulated content. The relevant transparency provisions apply from 2 August 2026.

Visible labelling may be used in particular where:

• a text was predominantly or entirely generated by an AI system and published without substantive editorial revision;

• no sufficient verification of facts or sources took place;

• a realistic image was artificially generated and could be mistaken for an authentic photograph;

• an existing recording was materially altered;

• an artificially generated voice imitates a real person;

• a video creates a false impression of a real event or statement;

• the technical method of creation is relevant to the proper interpretation of the content;

• Léoré considers voluntary labelling appropriate in the interest of enhanced transparency.

Separate labelling is generally not required where:

• AI is used solely for spelling, grammar or ordinary linguistic editing;

• a text written by a human is linguistically improved;

• AI is used for translation and the translation is reviewed by a human;

• an AI-generated draft is substantively reviewed, edited and assumed as the responsibility of the editorial team;

• AI is used solely for internal idea generation or structuring;

• proprietary image, audio or video material is only technically optimised without misleading substantive alteration;

• technical assistance has no relevant impact on the meaning or authenticity of the final content.

Each case is assessed individually.

9. Form and Placement of Labels

Required labels are presented clearly, comprehensibly and in a manner perceptible to the relevant audience.

Depending on the content, the following notices may be used.

Predominantly AI-generated texts

Transparency notice: This article was predominantly created using an AI system and was published without comprehensive human editorial revision.

AI-generated images

AI notice: This image was generated using artificial intelligence.

Materially altered images

AI notice: This image was materially altered using artificial intelligence.

Audio or video recordings

AI notice: This recording contains artificially generated or materially manipulated visual or audio elements.

Illustrative content

Illustration: This image was created using an AI system and does not constitute documentary material.

The wording may be adapted to the medium and context.

Where visual, audio or video content could be misleading, the label is placed so that it is recognisable when the content is perceived. A notice that is only hidden in technical metadata or on a separate general information page may not be sufficient for a specific item of content.

10. Interviews, Quotations and Personal Statements

AI-supported tools may be used to transcribe, structure or linguistically edit interviews.

The following principles apply:

• the meaning of a statement must not be altered through automated processing;

• direct quotations must not be invented or supplemented without a factual basis;

• statements must not be attributed to a person without a verifiable basis;

• editorial shortening must not misleadingly alter the original context;

• linguistic smoothing is carried out with particular care;

• agreed approval procedures are respected;

• confidential statements are not made public solely because they were technically processed;

• interview data is processed only in accordance with the relevant legal and contractual requirements.

Where interviewees are given an opportunity to approve quotations or review agreed passages, overall editorial responsibility nevertheless remains with Léoré.

11. Personality Rights and Protection against Deception

Léoré does not use artificial intelligence for the purpose of attributing statements, actions or characteristics to persons that do not correspond to reality.

Without an appropriate legal basis or consent, Léoré does not intend to publish artificial representations that:

• realistically imitate a real person in a deceptive manner;

• create the impression that a person made a statement they did not make;

• place a person in a fabricated context;

• impair a person’s dignity, reputation or personality rights;

• present an event that did not occur as authentic documentation.

For satirical, artistic, fictional or illustrative content, Léoré considers whether the artificial nature of the content is sufficiently apparent to the audience. An additional label is added where required.

12. External Authors and Partners

External authors, freelancers, service providers and cooperation partners may also use AI-supported tools.

Where content is published under the responsibility or brand of Léoré, external contributors may be required to:

• disclose relevant use of artificial intelligence;

• identify relevant sources in a transparent manner;

• respect personality rights and copyright;

• protect confidential information;

• apply legally required labels;

• comply with Léoré’s editorial standards.

A publication may be rejected or revised where:

• the origin of the content is not sufficiently transparent;

• facts or sources cannot be verified;

• AI-generated statements are presented as statements made by real people;

• third-party rights have not been adequately considered;

• a required label is missing;

• confidentiality or data-protection requirements have not been observed;

• Léoré’s quality standards are not met.

Paid partnerships and economic interests are labelled in accordance with the applicable requirements, irrespective of whether artificial intelligence was used.

13. No Fully Automated Editorial Operation

Léoré does not operate an editorial system in which AI independently selects topics, creates final content and publishes it without human responsibility.

Automated tools may support individual work processes. However, decisions on topic selection, context, editing and publication remain with a responsible person.

Léoré follows a human-centred approach:

Technology supports editorial work. Judgement, contextual understanding, diligence and responsibility remain human tasks.

14. Relationship to the Privacy Policy

These Transparency Principles explain the editorial and organisational use of artificial intelligence.

They do not replace Léoré’s Privacy Policy.

Where personal data is processed through the use of ChatGPT Business or another service, the relevant processing is assessed and documented in accordance with applicable data-protection law.

For all matters concerning:

• the identity of the data controller;

• processing purposes;

• legal bases;

• categories of personal data;

• recipients and service providers;

• international data transfers;

• retention and deletion periods;

• rights of data subjects;

• contact and complaint options,

only the current version of Léoré’s Privacy Policy shall apply.

Privacy Policy:
www.leore.de/datenschutz

15. Ongoing Development of Our Standards

The technical possibilities and legal requirements relating to artificial intelligence continue to evolve.

Léoré therefore reviews these Transparency Principles regularly and updates them where necessary.

In particular, we consider:

• legal developments;

• guidelines issued by competent authorities;

• journalistic and editorial standards;

• changes to the technical systems used;

• data-protection and security information provided by service providers;

• experience gained through editorial practice;

• substantiated feedback from readers, partners and affected persons.

The version currently published on this page is authoritative, subject to the language provision below.

16. Contact

Questions regarding our editorial standards, the use of ChatGPT Business, other AI-supported tools or a specific publication may be sent to:

Léoré Group
Email: info@leore.de
Website: www.leore.de

Where a question relates to a specific publication, please include the title and the relevant location of the content.

Legal Notice

These Transparency Principles describe the editorial and organisational practices of the Léoré Group when using artificial intelligence.

They do not constitute a privacy policy and do not provide general legal advice.

For all matters concerning the processing of personal data, only the current version of Léoré’s Privacy Policy shall apply.

All applicable statutory provisions remain authoritative, in particular:

• Regulation (EU) 2024/1689 laying down harmonised rules on artificial intelligence, also known as the EU Artificial Intelligence Act or EU AI Act;

• the General Data Protection Regulation;

• copyright law;

• media law;

• competition law;

• the statutory provisions protecting privacy and personality rights.

These Transparency Principles take into account the transparency requirements under Article 50 of the EU AI Act, which apply from 2 August 2026 and concern, among other matters, certain AI-generated or AI-manipulated content.

Authoritative Language Version

The German-language version of these Transparency Principles is the original and authoritative version.

The English and Japanese versions are provided for information and accessibility purposes. In the event of any discrepancy, ambiguity or difference in interpretation between the German version and a translated version, the German-language version shall prevail.