Credit reporting and data services
Experian’s generative AI guidelines, annotated
Experian’s brand site has seven rules for using AI, each with a do and a do not, from consent to disclosure. Quoted with notes, with what it leaves out.
Experian is a credit reporting company. It publishes its rules for generative AI on Brand Central, its public brand site. Seven numbered points cover text, images, video, and audio, and each point has a “Do” and a “Do not”. It is one of the fullest published examples of the using AI part of a guide. It also connects AI work to the rest of the guide: AI images follow the photography rules, and AI scripts follow the brand voice.
Best parts, quoted with notes
Photography and imagery
Ensure AI-generated images and videos follow Experian photography standards and are photorealistic
Has: where it applies. Holds AI images to the existing photography rules instead of writing a second set.
Voice and tone
Develop audio scripts or marketing materials using our Expedition Leader brand voice principles, ensuring delivery sounds natural and regionally appropriate
Has: exact value. Names the voice persona that AI-written scripts must follow. The persona has its own page in the guide.
Using AI
GenAI is a tool to support the creation of our work, it does not replace our judgment. If you create or commission GenAI content, you are responsible for ensuring it is accurate, appropriate, and on brand before it is used.
Has: where it applies. Puts responsibility on the person who made or commissioned the content, not on the tool.
Using AI
Prompt for copyrighted styles, characters, or distinctive brand aesthetics (for example, “in the style of Pixar”)
Has: right and wrong. Listed under 'Do not', with a real example of a prompt that breaks the rule.
Using AI
Obtain written consent before using the image or likeness of any identifiable person (including employees, clients, or members of the public)
Has: exact value, where it applies. Says consent must be written and that it covers employees too, not only the public.
Using AI
Include personal or regulated data (for example financial or health data) in prompts or uploads, even when using enterprise‑approved AI tools
Has: where it applies. Listed under 'Do not'. Says the ban still applies inside approved tools.
Using AI
Disclose use of GenAI where it could influence customer understanding, regulatory interpretation, or trust, for example, when depicting scenarios that may reasonably be interpreted as real
Has: where it applies. Says when to disclose, with an example. A separate rule says not to add disclaimers by default.
Using AI
Always use enterprise GenAI tools approved for use and follow the acceptable use standards. Public or non-approved tools must never be used for asset creation intended for external use.
Has: where it applies. Clear rule, but the list of approved tools is on an internal SharePoint page.
What it leaves out
- Does not name the approved tools. The page links to an internal SharePoint site for that list. Source
- Shows no date or version, so a reader cannot tell when the rules last changed. Source
What to take from it
- Hold AI-made work to the rules that the guide already has. Do not write a second set of rules.
- Write every ban as a “Do not” with an example, such as the prompt that breaks the rule.
- Name the approved tools where readers can see them, or say who to ask.