Gains:
- Ability to understand the place of generative visual artificial intelligence (tools that produce images from text) in the advertising workflow and prepare visual prompts and moodboard drafts
- Ability to plan visual variations according to brand identity, format and channel dimensions and use artificial intelligence at the draft and concept stages
- Understand that it is necessary to verify copyright, personality rights, brand consistency and reality/ethical boundaries in productive visuals.
Advertising often appeals to the eyes first and then the ears. The visual of a poster, the first frame of a social media ad, the opening scene of a video; these are perceived before the message. In recent years, generative visual artificial intelligence (tools that produce images from text; you write a recipe, the tool produces the appropriate image) has radically accelerated advertising production. It is now possible to visualize a concept in minutes, try dozens of variations, and prepare a moodboard (visual direction board). But this power brings with it two of the most sensitive legal and ethical areas of advertising: copyright and authenticity.
In this unit, you will learn where and how to use generative visual AI in the advertising workflow; We will see where you should stand. The basic framework from the beginning: Generative visual AI is the tool of the concept and draft phase. Every visual that will be published cannot be used without being checked in terms of copyright, personality rights, brand consistency and authenticity. No matter how impressive an image is, if it imitates another work, shows a real person without permission, or misleads the consumer, that impressiveness is a risk.
The right place of productive visual in the workflow
The safest and most efficient use of visual AI is in the early stages: idea visualization, moodboard, composition experimentation, style search. The images produced at this stage are for internal use; It is not broadcast to anyone, it only serves to find the direction. For example, you can quickly test the question "Is the atmosphere of this campaign warm or cold; urban or natural" with ten visual variations. The risk is putting these draft images directly into the publication.
Writing visual prompts is the skill of this job. A good visual guideline includes: subject matter (what is it), style (photography or illustration, what atmosphere), composition (close-up or wide-angle), light and color, and branding elements. But calling it "produce" a brand logo, a real celebrity, or a copyrighted character is risky; If the vehicle imitates them, there will be a violation of rights.
The following table summarizes where visual AI is safe and where it is risky:
Usage
Security
Why
Moodboard / style quest (internal use)
safe
Not published, useful for navigation
Composition/concept sketch
safe
Then professional production is done
"Making up" the actual image of the product
risky
May mislead the consumer
Generate real person/celebrity face
high risk
Personal rights, permission issue
Imitation of another brand/copyrighted work
high risk
Copyright and trademark infringement
Final image going live (unmoderated)
risky
Copyright/authentication verification required
Tip: Position the generative image as a “reference and sketch.” Produce the final broadcast image with either original footage/design or clarify the commercial use and copyright terms of the tool you use. Do not make commercial broadcasts without reading the license terms of the tool.
Reality and not misleading the consumer
The image shown in the advertisement carries a promise. The image in an advertisement for a hamburger should reasonably correspond to the product being sold. Producing a product image that is "much better than it is" with generative AI and presenting it as if it were the real product may fall within the scope of misleading advertising. Likewise, it is deceptive to portray an AI-generated “user” or “expert” image as a real customer or authority. An increasingly common principle is transparency: in some contexts and platforms, AI-generated content is expected or required to be marked as such.
Step by step: from visual concept to publication
Step 1 — Brief and brand identity. Color palette, typography, atmosphere, dos/don'ts. Brand identity is the framework for visual instruction.
Step 2 — Create moodboard. Extract style and atmosphere variations with AI (interior use).
Step 3 — Select and clarify concept. Determine direction with human judgment.
Step 4 — Format and channel measurements. Apply the same concept to channel dimensions (square, vertical, horizontal).
Step 5 — Copyright, copyright, reality check. Mandatory check before publication.
three mini cases
Case 1 — Moodboard speed. An agency was going to present a client with 3 different visual aspects for a cosmetics launch, but photography was expensive. With generative AI, they prepared moodboards in three different atmospheres (minimalist, luxury, natural) and presented them to the customers. The client chose the "natural" direction; However, the final advertising images were produced with professional shooting. AI found the direction quickly, the broadcast image came out with real production.
Case 2 — Copyright conflict. A brand used an image produced with generative AI and bearing the distinct style of a well-known artist in an advertisement. The artist's representatives objected, the campaign was removed and the brand issued an apology. Lesson: an image that closely imitates a style does not avoid copyright risk by saying "AI produced it"; Pre-publication similarity checking is a must.
Case 3 — Reality limit. A food brand used generative AI to produce a product image that looks much fuller and more perfect than the real product and used it in packaging and advertising. Consumer complaints came and it was determined that the product did not match the image. Lesson: the advertising image should reasonably match the product; Exaggerated AI visuals are a risk of misleading advertising.
Four copyable templates
1) Visual instruction skeleton (for moodboard):
Purpose: moodboard for a [product] campaign (internal use, will not be published). Brand identity: color [palette], atmosphere [warm/cold], style [photo/illustration]. Task: Write visual prompt text for the same product in 4 different atmospheres: each with subject, style, composition, lighting and color. DO NOT use a real person, brand logo or copyrighted character reference.
2) Brand identity framework:
Brand visual identity: main colors [...], avoided colors [...], typography feel [...], do's [...], don'ts [...].Task: Create a "do/don't" checklist for visual production in line with this identity; Write how to check each image produced with this list.
3) Channel format adapter:
My visual concept: "[concept recipe]". Task: Write instruction notes to adapt this concept to the following formats: Instagram square (1:1), Instagram story portrait (9:16), web banner landscape (16:9). Specify what should stand out and what should not be cropped in each format.
4) Pre-publication copyright/authenticity checklist:
There is an AI-generated image that I am considering publishing. Task: I have a checklist to check before publishing: (1) excessive similarity to another work/artist's style, (2) recognizable real person/celebrity, (3) other brand logo/packaging, (4) overlap with the reality of the product (risk of misleading), (5) need to indicate that it is AI-made. Explain how to check for each item; You cannot do an exact copyright search.
Weak prompt / Strong prompt
Weak prompt:
Create a realistic advertising image with a famous actor holding our product.
This request carries the risk of both violation of personal rights (real person without permission) and misleading; should never be used.
Powerful prompt:
Purpose: moodboard for a sports drink campaign (internal use). Atmosphere: energetic, morning, open air; style: live photographic. Subject: unidentified, unknown person drinking water after running; wide angle, not close-up; natural light.Task: Write instructions for 3 variations with this recipe.Do not use a real/recognizable person, brand logo, or copyrighted element.Note that these images will be copyrighted and fact-checked before publication.
The second claim does not contain any recognizable persons or copyrighted elements, is designed for internal use and provides for auditing.
Common mistakes
- Putting the draft image directly into the publication. It is a productive visual concept tool; Pre-publication moderation is required.
- Generating real person/celebrity face. It is a violation of personal rights without permission.
- Imitating the style of an artist. Saying "AI produced it" does not eliminate copyright risk.
- Making the product look better than it is. If the visual and product do not match, it will be misleading advertising.
- Not reading the license terms. Broadcasting without knowing the commercial use and copyright terms of the tool is risky.
- Presenting AI production as real. Ignoring the expectation of transparency is an ethical and, increasingly, legal problem.
Attention: The defense of "AI produced it, it is not my responsibility" does not protect you legally. The responsibility for copyright, personality rights and authenticity of the published image belongs to the brand and the advertiser.
In summary
Generative visual AI provides tremendous speed in the early stages of the advertising workflow (moodboard, concept, essay testing). But its power is intertwined with sensitive areas such as copyright, personal rights and reality. Safe use, positioning AI as a blueprint and reference tool; The aim is to produce the final broadcast images with original production or to clarify the license terms. Every image that goes to publication; It must be inspected to indicate that it does not violate any other work/brand/personal rights, that it reasonably matches the product, and that it is an AI production, if necessary. Responsibility cannot be transferred to AI.
Application task
Choose a product. (1) Write down the brand visual identity (color, atmosphere, do/don't do). (2) Produce 4 moodboard guidelines for internal use with “visual instruction skeleton”; There should be no real person/logo/copyrighted element in any of them. (3) Choose a concept and adapt it to three different rates with the "channel format adapter". (4) Take out the “pre-publication checklist” and perform five checks one by one for a sample image and note it down. (5) Research the commercial license requirement of the vehicle you intend to use and summarize it in one sentence.
checklist
- [ ] I used the generative image at the concept/draft stage, I did not publish it unsupervised.
- [ ] I did not use real people, celebrities or copyrighted elements in the images.
- [ ] I checked for extreme similarity to another artist/brand's style.
- [ ] I have verified that the image reasonably corresponds to the product.
- [ ] I have clarified the commercial use and copyright terms of the tool.
- [ ] I considered the need to indicate that there is AI production where necessary.