Do we need to disclose AI-generated images?

AI-generated images should be disclosed when people could reasonably mistake them for real photography, real people or real events. Transparency helps audiences understand the production method and protects trust when imagery is highly realistic. Disclosure is especially important where synthetic media could affect how people interpret a claim, story or person.

Can AI-generated images show people from under-represented groups?

AI-generated images can depict under-represented groups, but they need careful human review because AI can reproduce bias from its training data. Synthetic representation can also be a poor substitute for showing real people and lived experience, particularly where communities are already under-represented. Use real imagery when authenticity depends on the actual people affected by the story.

How do we avoid deepfake risk in marketing images?

Avoid using AI images that impersonate identifiable people or could be understood as evidence of a real event, endorsement or statement. Review realistic images for likeness misuse, misleading context and potential fraud before publication. Deepfake risk should be part of brand safety and cybersecurity planning, not a last-minute creative check.

When should we use real photography instead of AI images?

Use real photography when the message depends on genuine customer experience, employee voice, community participation or lived experience. Real people telling their own stories can carry a credibility that a polished synthetic scene cannot. AI imagery is better suited to illustrating concepts or scenarios that do not claim to document reality.

How do we use AI images without misleading people?

AI imagery works best as a clear creative tool, not a substitute for truth

AI-generated images can help teams create original visuals for scenarios that are difficult to source through stock photography, particularly in complex fields such as healthcare and manufacturing. The line is crossed when a synthetic image encourages people to believe that a person, place, event or customer story is real when it is not. Clear AI disclosure protects brand integrity by helping audiences understand what they are seeing. Transparency matters especially when imagery is realistic enough to be confused with photography or video. AI images should support an honest message, rather than decorate an evidence-backed claim with invented proof. A strong creative idea can still be bold, playful and emotionally resonant without presenting fiction as fact.

Sources: Powered by imagination: how I’m making stock imagery creative with AI, Faking bad: why marketers need to maintain standards in a post-truth world

Human review keeps synthetic media connected to real people and real standards

Human-in-the-loop review is the practical safeguard for AI images because image generators can create convincing output without understanding truth, context or consequence. Review each image for accuracy, implied claims and whether it could be mistaken for a real person or real-world situation. Check particularly carefully for deepfake risk, including imagery that could impersonate business leaders or make fraud harder to spot. Representation needs the same care. AI systems can reproduce bias in their training data, so images of minorities and people with disabilities need thoughtful human oversight. In some situations, real photography is the more credible choice, especially when the communication depends on lived experience or groups that are already under-represented.

Sources: Powered by imagination: how I’m making stock imagery creative with AI

Speed and personalisation should not turn into false intimacy or made-up evidence

AI image generation makes it easier to produce tailored visuals quickly, but convenience is not a reason to lower the standard of truth in advertising. Start with what the audience needs to understand, then decide whether an AI image clarifies the message or risks obscuring it. A realistic scene should never imply a product outcome, customer experience or human story that cannot be supported. Personalised imagery also needs a clear purpose: using AI to understand audiences and anticipate relevant needs is different from using automation to process people more efficiently. Brand safety depends on setting boundaries before a campaign goes live, including who approves imagery, when AI disclosure is needed and when a real image is the honest option.

Sources: Faking bad: why marketers need to maintain standards in a post-truth world, The Prompt: Why brand matters more than ever in an AI world, Powered by imagination: how I’m making stock imagery creative with AI

We use AI imagery to expand imagination, while keeping people accountable for the truth

We believe AI imagery should make creative work more possible, not less trustworthy. The useful question is not whether an image was made with AI, but whether it helps people understand something clearly without pretending to be a record of reality. Our work starts with the audience, the context and the claim before the visual idea takes shape. We have used AI-generated imagery for UST's AI services microsite, creating realistic scenarios for difficult-to-show sectors including healthcare and manufacturing. We also recognise the risks: deepfakes, likeness misuse and biased representation need human judgement, not a tick-box exercise. That is why we pair imaginative thinking with clear disclosure, careful review and an honest choice between synthetic visuals and real photography.