What should we disclose when we launch AI-made advertising?

AI disclosure should help an audience understand what they are seeing, not distract from an exaggerated or misleading message. Consider whether AI-generated or materially edited imagery, copy or video could create a false impression about a product, person, result or experience. If it could, the underlying execution needs review as well as any disclosure.

Transparency also needs human oversight. AI can enable personalised and interactive experiences, but it can also make manipulation, impersonation and biased representation easier. Review the creative for realistic but inaccurate imagery, misleading product implications, use of likenesses and assumptions embedded in the source material. The aim is not to label every use of AI automatically, but to ensure the audience is not left with an impression the brand cannot stand behind.

AI can be useful for research, early ideas and production support, but it does not reliably preserve a distinctive brand voice or the quality of thought that makes B2B content credible. Human editors should decide whether the language sounds like the organisation, whether its claims are proportionate and whether the work has something meaningful to say.

Set the brand behaviour before selecting tools, prompts or automated workflows. Brand behaviour covers how the organisation acts at every touchpoint, including non-human ones, rather than treating tone of voice as a layer added after the technology is built. This creates a clearer basis for deciding where AI should be direct and transactional, where it should hand over to people, and where it should not be used at all.

Emotional storytelling can be effective when it helps an audience understand a real issue, but bold creative should not be designed for attention alone. Start with what the audience likes, wants or needs, then test whether the message is engaging without being deceptive, manipulative or out of character for the brand.

Use direct audience and stakeholder input alongside desk research. Interviews can reveal the practical concerns, language and expectations that a campaign must address, especially where technical subject matter is being translated for broader business audiences. Review proposed AI-assisted executions with the same care: ask whether the creative reflects the audience's reality, whether the emotional device is appropriate, and whether the promise remains clear when separated from its most dramatic imagery.

A claim is stronger when its scope, evidence and limitations are understood before the campaign is produced. This matters for product performance, sustainability and purpose-led communications, where a broad statement can imply more than the available proof supports. Keep the claim aligned with the organisation's actual behaviour, not only its preferred positioning.

Build a record that identifies the source, methodology, date, scope, owner and approval for each objective claim. For environmental messaging, include material context in the advertising itself rather than relying on a qualification that audiences may not see. Connecting communication to wider actions also makes the message more credible: brand behaviour is not limited to words, and a campaign is more persuasive when its promise is reflected across the customer experience.