Transparent Use of AI in Higher Education Communications and Marketing

The way we work in communications and marketing has fundamentally changed due to the availability of generative AI applications. The Communications and Marketing Office at the University of Kassel harnesses the potential of these technologies responsibly, guided by the Journalistic Guidelines and Recommendations for the Use of Generative AI in University Newsrooms issued by the Federal Association for University Communications.

The University of Kassel uses generative AI responsibly in its editorial work

We adhere to the EU Code of Conduct on Transparency of AI-Generated Content and have derived the following principles for the University of Kassel from it:

  • Authentic, Credible and Respectful
    We use AI in a way that ensures the University of Kassel, its members, facilities, and programs are portrayed in a factually accurate, respectful, credible, and authentic manner.
  • Reviewedby humans
    We carefully review all content created or edited using AI before publication. In doing so, we pay particular attention to factual accuracy and to avoiding discriminatory content and stereotypes. Academic content is reviewed by appropriately qualified staff members.
  • Respecting Data Protection
    As a general rule, we enter personal and confidential information into AI applications only in anonymized form. For non-anonymized data, we use exclusively applications approved by the university that comply with data protection regulations.
  • Protecting personal rights
    We process photos and videos of individuals using AI only if explicit consent has been given. 

Legal Framework for the Use of AI

The legal framework is provided by the European Union’s AI Regulation, known as the AI Act. Among other things, it specifies when AI-generated or AI-modified content must be labeled. 

Labeling is required in particular when:

  • image, audio, or video content has been artificially generated or altered in such a way that it can be perceived as genuine (“so-called deepfakes”),
  • published texts have been largely generated or substantially altered using AI and serve to inform the public about matters of public interest.

Labeling is generally not required if 

  • AI is used solely for support purposes, such as for linguistic revisions or translations,
  • media are technically edited without substantially altering their message,
  • AI-generated or AI-modified texts have undergone human review or editorial oversight prior to publication, and a natural or legal person bears editorial responsibility for the publication.

For artificially generated or altered image, audio, and video content that qualifies as a “deepfake,” the requirement for labeling through human or editorial oversight generally remains in effect.

Our AI labels and what they mean:

This icon indicates website content that has been modified using AI—for example, by replacing people, adding or removing realistic image elements, or rewriting large sections of text.

Download the icon

This icon identifies website content that was created entirely or almost entirely using AI, such as a computer-generated image, a synthetic audio recording, an explanatory video, or text written largely by AI.

Download the icon

Your cookie settings have blocked this video

The label indicates that content has been created or modified using AI. However, responsible use of AI goes beyond this: content must be fact-checked, legally permissible, and consistent with the university’s values.

Specific Labeling Cases 

If you're unsure when your content must be labeled as AI-generated or AI-modified, we offer guidance here to help you decide.

Image: KPs Photography - stock.adobe.com

AI Support for Text Editing and Writing

No labeling required

  • Check grammar
  • Polish the text
  • Translate and then review

Attribution is generally required

  • AI creates a first draft; the editorial team revises it and is responsible for the final text. This applies in particular when AI performs significant creative work or generates science-related content.

Disclosure required

  • AI analyzes and summarizes scholarly texts that editors are unable to evaluate from a technical standpoint.
  • An article generated entirely by AI is published virtually unchanged.
Image: ARAMYAN - stock.adobe.com

AI Support for Image Editing and Image Creation

No labeling required

  • Remove noise
  • Sharpen image
  • Correct colors
  • Optimize exposure and contrast
  • Slightly blur the background

Labeling is generally not necessary

  • Completely remove the background

Labeling is generally necessary

  • Replace the background

Marking is necessary

  • Add or remove people
  • Generate or add new objects
  • Fill in missing image areas
  • Use a fully AI-generated image
Image: Aon Khanisorn - stock.adobe.com

AI Support for Video Editing and Video Production

No labeling required

  • Reduce image noise
  • Stabilize image
  • Optimize brightness, contrast, or colors
  • Sharpen or upscale the image
  • Generate automatic subtitles
  • Remove background noise
  • Normalize volume
  • Automatically transcribe spoken content
  • Automatically cut scenes or select highlights
  • Make minor retouches, such as reducing skin blemishes

Tagging is generally necessary

  • Replace the background, for example with a virtual studio

Marking required

  • Remove objects from the video
  • Add objects or people
  • Adapt lip movements to a new language
  • Synthetically replicate the voice of a real person—deepfake
  • Make a real person’s avatar speak – Deepfake
  • Create a fully AI-generated explainer video with no reference to real people
  • Create an AI-generated recording of a real person
  • Create an AI-generated scene of an event that never happened or a real event

Scope of Application

These rules apply to all content published by the university - including both digital content and print materials such as posters, flyers, and brochures.