Where does the harm come from?
An AI-generated visual is as neutral as a photograph; what harms a brand is how and where it's used. We see the risks under three headings:
- Misleading people: presenting a hotel room that doesn't exist, a feature the product lacks, or a customer experience that never happened as if it were real.
- Identity drift: colour, light and typography sliding into a different tool's style with each visual, away from the brand guidelines.
- Use without consent: using the face or voice of an employee, a spokesperson or a well-known person without their explicit permission.
Our principle is clear: we don't produce content that misleads consumers, and we don't recommend it to our clients.
What are synthetic media and digital twins?
Synthetic media are images, video and audio created or transformed by AI; a digital twin is a model built from recordings of a real person, carrying their likeness and voice. A person's likeness and voice belong to them: a digital twin is used only with their knowledge and consent, and AI use is stated clearly where viewers need to know.
When is synthetic media the right choice?
They help when they carry a message to more people without changing the truth:
- Carrying a message the person really gave into other languages or formats.
- Building an abstract, fictional or imagined world for the brand.
Where it isn't right: making someone appear to say something they didn't, re-enacting a real customer experience or testimonial, or changing the truth of a product or a place.
When is a real shoot essential?
- Selling the product itself: colour, texture and size must match reality exactly.
- Presenting a place: a hotel, restaurant, factory or office should look as it is; if there are no real photographs, our production team shoots them.
- Portraying real people: customer testimonials, your team and your events are told through real footage.
- In a regulated sector: in fields such as healthcare, we recommend consulting legal experts in that sector.
Where does AI genuinely help?
- Concept and mood: making ideas visible before the shoot.
- Variations: adapting an approved visual to different formats and platforms.
- Imagined scenes: building abstract or fictional scenes that couldn't be filmed in reality.
For Soulqin we produced films and social media posts with the help of AI; the test was to fit the brand's identity and not mislead about the product.
Which questions should you ask before publishing?
- Could viewers take this image to be real? If so, does what it shows actually exist?
- Does it follow the brand guidelines: colour, light, typography, tone?
- If a recognisable person appears, have they agreed to it?
- Does it echo a well-known work, a real person or another brand?
- Are there technical flaws? Hands, lettering and small details are easy to miss.
If any answer is unclear, fix the visual or plan a real shoot instead of publishing.
How do you protect your brand identity?
Consistency comes from guidelines, not tools: when the logo, colours, typography and corporate identity guide are clear, every visual looks like the same brand. So the visual identity must be solid before production; when a real shoot is needed, our photo and film production team steps in.