Producing ad images, product photos, video and even digital characters with AI has become both a fast and a cheap path for small businesses. At the same time there is movement on the legislation side: according to news reports, the Ministry of Trade’s regulation amendment on advertising and unfair commercial practices brought onto the agenda the clear labeling of AI characters that are hard to distinguish from humans, and a ban on the misleading use of digital copies of real people. The amendment was reported to have come into force on August 1, 2026.
This article explains under which headings a business should evaluate the content it has, in which cases it should think about disclosure and labels, and how to set up a simple internal routine. An important warning: this article is not legal advice. News sources use different wording on details; read the exact text in the Official Gazette and the Ministry of Trade announcements, and make your application decisions with your legal adviser. We described the production side of producing ad creative with AI in our AI ad creative post.
What Do the News Reports Say?
According to the summaries in the press, the headings of the regulation amendment related to AI and labeling are these:
- It is written that AI characters used in advertising and indistinguishable from humans must be stated clearly, understandably and distinguishably.
- It is stated that the use of digital copies of real people created with AI in a way that misleads the consumer was prohibited.
- Another source writes that images, audio and video produced with AI cannot be deceptive or create an unreal perception for the consumer, and does not state a separate requirement for labeling these contents.
- It is explained that in influencer posts, an “advertisement” or “promotion” label is required if there is a benefit in return.
- There are rules against fake consumer reviews and against selecting reviews for publication.
Note the difference between sources: one stresses that a specific label is needed, the other that the principle of not misleading comes to the fore. So the practical approach is this: AI-generated content is risky if it makes the viewer believe something unreal; when in doubt, stating it clearly is safer. The final answer on which label is mandatory in which case comes from the official text and your legal adviser.
First, Create a Content Inventory
The first step of preparation is knowing what AI use you have. In many businesses AI use is scattered: an employee produces the image, an agency prepares the video, an influencer uses their own tool. Without knowing which content is used where, risk cannot be managed.
Create a table with these headings: content type (image, video, audio, text, character), where it is used (ad, website, social media, email), the role of AI (fully generated, edited, voice cloning, face swap, background change), who produced it (employee, agency, influencer), whether it represents a real person and whether it changes the real appearance of the product.
This table can be divided into three classes: low risk (text drafts, background cleanup, decorative images), medium risk (adding elements to a product image, animation, synthetic voice) and high risk (a real person’s appearance, fake customer review or testimonial, an image that misleads about the product’s real feature, a virtual “expert” or “doctor”). Set up a separate approval step for high-risk uses. For an internal policy you can see our AI use rule draft.
AI in Product Images: Where Is the Limit?
AI in product images is the most common area and the one with the most mistakes. Operations like background cleanup, lighting correction or placing the product in a different setting (mockup) usually do not cause problems; but operations that change the real features of the product can.
Watch for these: do the product’s color, size, texture and material match the real product? Is a feature that is not on the product (an extra part, accessory, different model) in the image? Can the “usage example” image really be made with this product? If something that does not come with the product (a “gift”) is shown, is it clearly stated?
A significant part of customer returns and complaints comes from the image not resembling the real product. An AI image can widen this gap. The safe path: let the main images on the product page be photos of the real product; show AI-generated lifestyle images in a separate area and, where needed, add a clear note like “image is illustrative”. Also review default features such as automatic retouching of product images on ad platforms; for details see our Threads and Advantage+ post.
Digital Characters, Virtual Presenters and Voice Cloning
An AI-generated presenter or brand character is a consistent and low-cost way of communicating for some businesses. But the rule that came to the agenda in the news targets exactly the clear disclosure of such characters that are hard to distinguish from humans.
Practical suggestions: if a digital character used in an ad looks like a real person, use a clear statement such as “produced with artificial intelligence” in a visible place in the video and in the description; do not make the statement small and unreadable. Do not copy the image or voice of a real person (employee, expert, celebrity, customer) with AI; that person’s explicit written permission is needed and misleading use may also be prohibited. Avoid giving advice through virtual characters with titles such as “doctor”, “lawyer” or “expert”; the news also mentions a ban on the misleading use of academic titles. Never use AI content that looks like a customer testimonial or review; that means a fake review.
The same principle applies to voice cloning: imitating a person’s voice without permission creates legal and ethical problems. If you use a synthetic voice in an audio ad, use it in a way that does not mislead the listener, with a disclosure if needed.
Influencer Content: Advertising Label and AI Use
Influencer collaborations involve two separate topics: the advertising label and AI use. According to the news, if there is a benefit in return, the post must be marked as “advertisement” or “promotion”. Because in-kind benefits such as a free product can also fall within this scope, keep your risk area wide.
Write these into the collaboration contract: the advertising label must be in a visible place in the post; prior approval is needed if AI-generated images, audio or face changes will be used in the content; no unverifiable claims about the product; content presented as real customer experience must actually have been lived; content is shown to the brand before publication.
If you want to use a virtual influencer produced with AI (a person who does not exist), it is safer to state clearly that this person is an AI product. “Face swapping” a real influencer with AI or presenting it to the consumer as real experience is high risk. For types and ways of working see our influencer post and our nano influencer post; for other changes on reviews and discounts read our e-commerce rules post.
Platforms’ Own Labels and Rules
Besides legislation, ad platforms have their own rules. Meta, Google, TikTok and other platforms can offer labeling tools or guidelines for AI-generated content; guidelines change over time. If the platform has its own labeling option, use it; if not, state it yourself in the content or description.
Platform policies also affect ad approval: misleading content, content with unreal claims, content showing real people without permission or manipulative content can be rejected. Common rejection reasons for AI-generated images include broken anatomy, unreadable fake text, unreal logos and images containing medical or financial claims.
Check your ad account’s policy warnings regularly. An account restriction is far more expensive than the rejection of a single ad. Follow the help pages of the relevant platform for account health; the exact rules on this vary from platform to platform.
An 8-Step Internal Policy Draft
Instead of leaving AI use without rules, write a short internal policy. The 8 steps below give a start for a small business.
- Keep a content inventory. Write into a table which content uses AI, who produced it and where it was published.
- Set risk classes. Write low, medium and high risk examples; put an approval step for high-risk use.
- Set a real-person rule. Do not copy a real person’s image or voice with AI; get written permission if needed.
- Protect product reality. Let the product’s color, size and features match the real product in the image.
- Write a label rule. Use clear disclosure in AI content that is hard to tell from reality.
- Avoid fake testimonials and reviews. Do not produce customer reviews or testimonials with AI.
- Add to agency and influencer contracts. Write the AI use, the advertising label and the approval process.
- Set up a record and review routine. Review published content once a quarter; update the policy as legislation and platform rules change.
However short the policy is, it should be written; verbal rules get lost as the team changes. For the general framework of AI marketing work see our AI marketing post.
Common Mistakes
Skipping disclosure by saying “everyone does it”. Rules and platform policies change; early compliance is cheaper than fixing later.
Making the product look better than it is. An unreal image creates both return and complaint risk and misleading advertising risk.
Using a real person’s image or voice without permission. The legal and reputation risk is high.
Producing fake reviews or testimonials. This is a clearly problematic practice.
Not knowing what the agency or influencer uses. Responsibility does not stay only with the producer; the brand must check too.
Treating a news summary as the official text. Compilations like this article do not replace the official text; work with your legal adviser.
How Does Soft Ajans Help on This Topic?
At Soft Ajans we run ad creative, video and image production, social media management and ad account setup. In AI-assisted production we take care to set up a workflow that protects product reality, is not exaggerated and is traceable; we do not produce content that shows a real person’s appearance or contains fake testimonials. We do not provide legal advice; your legal adviser should be the reference for compliance decisions. We do not guarantee sales or approval.
For details take a look at our production and advertising agency page and our social media management page, and you can request a free preliminary analysis.
Frequently Asked Questions
Do I have to label every image I produce with AI?
Sources use different wording on this. In the news, the disclosure of AI characters indistinguishable from humans stands out; the principle of not misleading the consumer applies to all content. For a definite answer look at the official text and your legal adviser; when in doubt, adding a disclosure is safer.
Can I change the background of my product photo with AI?
As long as it does not change the product’s real appearance it is generally low risk. Do not change the color, size or features; state that the image is illustrative if needed.
Can I produce an AI image of one of my real employees?
Do not do it without written and explicit permission. The misleading use of digital copies of real people is mentioned in the news among the practices that were prohibited.
Can I use AI content that looks like a customer review?
No. This means a fake review and testimonial and is risky under both legislation and platform rules.
When is an advertising label needed in an influencer post?
According to the news, when there is a benefit in return. A free product can also be a benefit; ask your legal adviser and write the label requirement into the collaboration contract.
Where should I start as a small business?
Make a content inventory, determine risk classes and write a short 8-point internal policy. Add the AI use and advertising label items to agency and influencer contracts.
