AI Use and Transparency: Our Obligations

AI is opening up new possibilities for Bayer communicators, from creating content and visuals to accelerating workflows and exploring creative ideas. As AI becomes part of our everyday work, transparency requirements apply to how we create and publish content. This page explains the rules, responsibilities, and practical steps needed to meet these requirements and use AI in a compliant, transparent, and trustworthy way.

Many colleagues across Bayer’s brand and communication teams are already using generative AI to accelerate output, make workflows more efficient and create content that is aligned with our mission and brand identity.

What we produce goes beyond text. Many AI engines – whether they are general-purpose platforms like ChatGPT or Gemini, or Bayer’s own Media Pool GenAI, support multimedia outputs. These could include:

• Images e.g. visuals, creative concepts or photorealistic imagery

• UI User interface elements

• Video

• Audio e.g. voiceovers or podcasts

• Strategy and brief documents

• Creative copy

Using AI opens up more possibilities for us – but it also gives us more responsibilities to follow. In general, these fall into two categories: 

• Using AI only where it is appropriate (i.e. safe, legal and rights-compliant)

• Being transparent about our use of AI and labelling generated content.

These obligations are outlined in Bayer’s Principles for Responsible AI, Governance Procedure – Compliant Use of AI,  Bayer AI Transparency Guidance  and AI Compass for Employees. This page will give a summary of the practices that are most relevant to branding use cases.

Why we have AI safety and transparency policies

When we use any new technology, we do so in a way that fits both with regulations and with our values and personality. AI is no different. There are three main reasons for giving clear AI guidance:

• Legal requirements: AI is an increasingly regulated area. In particular, we all must adhere to the EU AI Act. New transparency requirements within this act (Article 50) came into effect in August 2026. Fines reach up to €15M or 3% global annual turnover for breaching these requirements and rules

• Image-related rights: When a photograph is taken, rights are automatically assigned to the photographer and to any people depicted. For images useable by Bayer, creators and subjects have signed releases that permit our use – but many of these releases pre-date AI, and therefore may not cover uploading to AI engines or editing with AI. Altering recognizably real photos and people using AI may violate their rights.

• Values and respect: We are a people-oriented organization. We communicate with warmth and empathy. Therefore, any AI-generated content should be fair, positive and supportive towards its depicted subjects – not derogatory or stereotypical.

When generative AI use is and isn’t permitted (images and video)

Examples

Original	AI edit	Acceptable
		Yes Original is a Bayer-owned or licensed image with Model Release agreement signed. AI edit makes the characters unrecognizable.
		No One of the subjects remains the same, making the image recognizable. This means modifying the other subject’s traits is not permitted. We may also be violating the photographer’s rights here.
		Yes This time, both characters are replaced, creating an original and compliant result.

In fact, it is not just people you need to take care to represent properly – but backgrounds and environments too.

Examples

Original	AI edit	Acceptable
		No The environment has been modified but the people depicted are the same. The photo is therefore recognizable, and our edit may infringe the subjects’ or photographer’s rights and erode trust in us.
		Yes This time, both the people and the environment are changed. This is an acceptable use.

Broadly, the same rules apply to video. You can produce videos depicting AI-generated characters, including photorealistic ones, as long as they are fully distinct and not based on recognizable people (and we have the required usage rights for any source assets). Likewise, AI voiceovers are permitted as long as they are not based on, or mimicking, the voice of an identifiable person.

Just as with images, AI-generated or AI-modified videos must be labelled. We will cover how to do that in the next section.

Transparency requirements: when to label AI use (including image, video, audio)

If we have used AI or are delivering AI-generated content to users, we have an obligation to disclose it. This applies to AI systems we have developed ourselves, but also when we deploy third-party AI tools – which means we need to label our AI-generated or AI-modified content in branded applications.

What about content sent to us by agencies or other third parties?

Agency-supplied content also needs to follow our labeling requirements as described in AI Transparency Guidance. We should require agencies to label AI-generated content; we have prepared a legal text that you can use to stipulate this.

We should not remove any labels or disclaimers indicating AI use from agency-submitted content.

If their label is a different style to Bayer’s, this is acceptable as long as it meets our standards for clarity. However, if their labelling is unclear, badly placed or otherwise does not meet our standards, we should correct it or replace with our own labels.

We all need to follow the guidance in this article – however, it is not designed to discourage you from using generative AI. On the contrary: you are absolutely permitted and encouraged to use it. Remember to do so in a legal, compliant and transparent way, using this article to guide you and give you confidence.

One more thing to remember: Bayer’s communications standards and values apply to AI-generated content as much as anything else. So, as well as these AI safeguards, be sure to consult the branding guidance explored elsewhere in the Identity Net.