Transparency when using AI: how much is enough?
Hiding the role of AI can undermine trust; excessive labeling devalues craftsmanship. Where is the line?
Sinh Vũ uses AI as a tool in the process and transparently communicates this according to the nature of each task: content released externally needs to be clearer than internal drafts, human likeness or voice needs to be labeled, while AI supports ideas where the final product is selected by a person, transparency at the process level is sufficient. The core principle: clients have the right to understand when a product has AI involvement to trust and provide feedback appropriately.
- AI significantly creates or edits published content
- create human figures with human voices or testimonials
- the market has labeling laws or is sensitive regarding trust
- label AI on everything, even in places where it’s unnecessary
- AI only assists with draft product ideas finalized by humans
- Present AI content as a genuine customer experience.
The question Sinh Vũ receives more frequently now is: should we disclose the use of AI in brand design, how should we say it, and how much is enough? The answer is neither to "hide it" nor to "label everything." The appropriate level of transparency depends on the type of product, the extent of tool intervention, and the expectations of the recipient.
Why transparency is not optional
Being transparent about AI is not a humble gesture or a marketing tactic. It is a duty of honesty: clients have the right to understand the process behind the products they are paying for, so they can trust and provide feedback appropriately.
Two foundational principles from international AI governance frameworks point to the same thing. The OECD AI Principles refer to this as transparency and explainability, meaning that AI systems and users of AI tools must be able to explain what happened during the result generation process. The NIST AI RMF emphasizes accountable and transparent as core attributes of responsibly deployed AI.
Using AI does not exempt you from the obligation of honesty. Presenting AI-generated content as genuine testimonials from customers without clarification is misleading, even if the original emotions are real.
FTC staff guidance on AI in advertising and endorsements, 2025
Two axes to define transparency
Not every output requires the same level of disclosure. Sinh Vũ uses two axes to determine.
Axis 1: type of output. The content released externally (articles, images, videos, testimonials) needs to be clearer than internal documents or draft ideas. Viewers of the published content do not have the opportunity to ask about the process, so the information must be included within the content or accompanying notes.
Axis 2: level of AI intervention. AI can create a complete image or text that is fundamentally different from AI-assisted ideas and drafts that professionals select and refine. The higher the level of intervention, the greater the need for transparency.
Transparency at the process level is sufficient when: AI only supports ideas, variations, and background research. The final product is the result of judgment and selection by professionals. In this case, transparency means being ready to explain when customers ask, not necessarily labeling every file proactively.
Common mistakes when addressing this topic
- Completely hide and let customers discover. This is the worst mistake. When customers realize not because they were told but because they found out, the loss of trust is much greater than if it had been stated directly from the beginning.
- Label "made by AI" on everything indiscriminately. This approach leads clients to think the studio just presses a button, diminishing the perceived value of expertise and the craftsmanship that truly exists in the product.
- Present AI-generated content as a real customer experience. While the original emotion is genuine, using AI to reconstruct testimonials without clearly stating it is a behavior that easily violates advertising rules, even in markets without specific laws.
- Thinking the Vietnamese market has no laws, so there's no need to worry: International standards are spreading rapidly, and large corporate clients are increasingly requiring AI transparency in service contracts.
The viewpoint of Sinh Vũ
Sinh Vũ positions AI as a tool to amplify expertise, not as a replacement. Transparency is understood in that spirit: the product the client receives is the result of the taste and selection of the professionals involved. AI is part of the process, not the author.
When communicating with clients, Sinh Vũ clearly states where AI accelerates (drafts, variations, foundational research) and where human decisions are involved (direction, selection, final edits, judgments aligned with the brand system). This distinction does not weaken the studio's position: on the contrary, it shows clients what they are paying for. Expert judgment anchored in the brand system is something AI cannot possess without human guidance.
In consulting sessions and workshops, transparency about the AI process is part of the value: clients gain a clearer understanding of the process, provide more accurate feedback, and trust the results more.
Decision checklist
Topic: Transparency when using AI with customers. Sinh Vũ guide, sinhvu.com
Select each item you find appropriate, then print or save as PDF to take with you.
If you have marked most of the signs above, this is the time to discuss in more detail. Sinh Vũ can help you review and propose a direction.
References
OECD AI Principles, Transparency and Explainability (Principle 7); NIST AI Risk Management Framework 1.0, section Accountable and Transparent; FTC staff guidance on AI in advertising and endorsements (2025); EU AI Act Article 50, transparency obligations.
Frequently asked questions
Vietnam's market currently lacks AI labeling laws; is transparency necessary?
The absence of rules does not mean there are no risks. International standards are spreading quickly, and large corporate clients are increasingly asking directly about AI processes. When clients discover issues later instead of being informed upfront, the loss of trust is much greater than if they were told from the beginning. Sinh Vũ recommends adopting international standards even when they are not yet mandatory.
If AI only helps draft and then the person completely rewrites it, should that be mentioned?
There is no need to label every detail, but you should be transparent about the process when clients ask or when signing a service contract. The final product is the result of the judgment and selection of the professional; AI is part of the process, not the author. That is the honest and sufficient answer.