
In short: Use AI where nobody is asked to believe a real person, place or product is on screen. Shoot for real when the viewer must trust what they see, when a product has to match the real thing, or when a real person speaks for your business. The ASA says advertisers stay responsible for AI output, so ask what your audience will assume they are watching.
Most business video needs a mix. Some parts can be generated quickly and cheaply, and some parts only work if a real camera has recorded a real thing. The common mistake is to choose by cost alone. Choose by what the viewer will believe they are seeing.
Ask these questions in order
Answer these for each scene. The first question you answer yes to decides the approach.
- Does the scene show a real person speaking as if they are real, such as a customer, a member of staff or a founder? If yes, shoot it.
- Does the scene show a real product or place a viewer might buy or visit? If yes, shoot it, or use a real photograph as the reference.
- Does the scene make a claim the viewer could check, such as a result, a number or a testimonial? If yes, shoot it and keep the evidence behind it.
- Is the scene a background, a motion graphic, a diagram or an illustration that nobody will read as evidence? If yes, AI can make it.
- Would your audience feel misled if they later learned it was generated? If yes, shoot it. If no, AI is a fair choice, and a person still checks the output.
Where the rules bite
The CAP Code covers ads in the UK, and the ASA applies it to AI-made content in the same way as any other content. Its June 2026 summary makes four points:
- The Code is media-neutral. Harmful, offensive or socially irresponsible AI images can breach it.
- A deepfake endorsement that looks real enough to mislead likely breaches the Code.
- The advertiser remains responsible for any output used in its ads. That includes checks for misleading claims, harmful or discriminatory stereotypes, and whether the content suits the audience and the placement.
- Automation does not shift responsibility. Ads that an AI platform generates or places still fall to the advertiser.
The ASA does not regulate personality rights or intellectual property. It advises taking legal advice before using a public figure's likeness without permission.
Two examples
- A product that must match. A bottle in a shampoo advert has to look like the bottle on the shelf. Shoot the product on a clean cove, then use AI for the motion graphics or the environment around it. Our infinity cove page explains how the set is built.
- A background for a pitch video. A softly lit office behind a real presenter can be generated, provided the presenter's claims are real and filmed.
Mistakes to avoid
- Using AI for testimonials. A face that belongs to nobody, presented as a customer, misleads the viewer.
- Shooting everything for real out of habit, and paying for a location the message does not need.
- Letting an AI tool choose the product's shape or colour without checking it against the real thing.
- Treating the output as finished. A person still has to check it before it is published.
For the stage by stage view of where AI fits in a production, read AI in the production workflow. The AI Studio page explains how we help a business work out where AI belongs.
Sources
- Advertising Standards Authority, AI and Deepfakes: Four Things Advertisers Need to Know Before They Hit Run (CAP news, 11 June 2026), read 5 November 2026
Related guides
- AI video for small businesses: what's real, what's risky
- AI in the production workflow, stage by stage
- Who is responsible for an AI-made ad?
- Voice and likeness: a consent and contract checklist
- How we clone a voice with consent
- All guides and tools
The picture at the top of this page is an illustration, not a photograph of a client or a real job.
Questions we get asked
When should a business use AI instead of filming?
Use AI for work where the viewer does not need to believe a real person, place or product is there, such as backgrounds, explainer graphics and draft versions. Film when the content makes a claim about a real thing, shows a real person or has to match the product exactly.
Can an AI-generated video be used in an advert?
Yes, provided it complies with the CAP Code. The ASA says the Code applies regardless of how content was created, and the advertiser remains responsible for the output used in its ads.
Do you have to say when a video is made with AI?
Ask whether a viewer would expect the content to be real. The ASA says a realistic AI endorsement that looks real enough to mislead likely breaches the Code.
granary.digital · 01243 884993 · studio@granary.digital · The Old Grainstore, Adsdean, Chichester, West Sussex, PO18 9DN · Prices exclude VAT unless marked inc VAT
