Generative AI has made it much easier to create convincing content. A well-produced photograph no longer tells us much about how that image was created. A well-written article may have started with a person, a machine or a combination of the two. Even someone speaking in a video may never have stood in front of a camera.
For brands, this creates a communications challenge that goes beyond whether consumers like or dislike AI. As authorship becomes less evident, audiences also lose some of the reference points they used to understand who shaped a message, whose judgement sits behind it and who is willing to take responsibility for what was published.
This issue is particularly relevant in Latin America, where perceptions of AI itself are relatively positive, but trust in the content people find online is far less straightforward.
In Brazil, 86% of participants in KPMG and the University of Melbourne’s 2025 global study on AI said they accepted or approved of AI. The figure was 80% in Colombia and Mexico and 79% in Argentina. At the same time, the same study found that 67% of Brazilians, 68% of Mexicans, 71% of Argentinians, 73% of Chileans and 75% of Colombians were unsure whether they could trust content found online because it might have been generated by AI.
That difference matters. These consumers are not necessarily rejecting the technology; they are becoming less certain about the information in front of them.
Accepting AI does not mean trusting everything it produces
Brazil illustrates this tension clearly, with more than half of participants in the KPMG study saying they were willing to trust AI and 91% expecting it to bring benefits, while 79% also expressed concern about possible negative consequences. In Colombia, the picture was similar: eight in ten respondents approved of AI and 91% expected benefits, but 75% expressed uncertainty about whether online content could be trusted when the technology might have been involved.
These two positions can coexist without contradiction. A person can use ChatGPT every day and still question an AI-generated testimonial, value an AI-powered customer service tool while also wanting to know whether the person appearing in an advertisement actually exists, and recognise the efficiency the technology brings to a company while still expecting that company to take responsibility for what it communicates.
For communications teams, this makes the divide between “AI versus human” less useful. The more relevant distinction is between using the technology as part of the production process and allowing it to make it less clear who is responsible for the final message.
This becomes even more important in an online environment where distrust is already growing.
Jumio’s 2025 study found that 76% of Mexican consumers had become more sceptical of the content they encountered online because of AI-powered fraud. Eighty-three percent were concerned about being deceived by manipulated content on social media, with similar levels of concern around AI-generated videos and voice impersonation.
The research deals specifically with fraud and manipulated content, rather than brand communications, but legitimate companies operate in that same environment. Consumers move between advertising, social media, news, influencer content and synthetic materials without necessarily making the same distinctions that communications professionals make internally.
As a result, some of the signals that once helped audiences assess credibility are becoming weaker. Production quality is no longer proof that significant human effort was involved, a realistic image is not necessarily a photograph, and a convincing voice no longer guarantees that the person actually said those words.
The challenge for brands, therefore, now has more to do with enabling audiences to understand who is behind the result.
Authorship matters more in some types of content than in others
That does not mean every use of AI should be treated in the same way.
A product description does not depend on human authorship in the same way as a CEO’s opinion. A background image creates a different expectation from a customer testimonial. A translated FAQ and a creator recommendation may both involve AI, but they serve fundamentally different purposes.
The more the value of a piece of content is tied to someone’s identity, judgement, expertise or experience, the more important authorship becomes.
Influencer marketing is a clear example. Part of the value of a recommendation comes from the perception that a real person is expressing a real opinion or describing an experience they actually had. AI can help edit, translate or adapt that content without changing its essential meaning, but if it starts manufacturing the experience or opinion itself, the nature of what the audience is being asked to trust changes.
The same applies to customer stories and expert commentary. A testimonial has value because a customer genuinely had a certain experience, and an article signed by an executive carries weight because it should reflect that executive’s perspective. If AI starts creating the very experience, opinion or expertise that gives the content its authority, the issue is no longer simply how the material was produced.
It becomes a question of provenance.
Transparency helps, but it does not replace human judgement
There is evidence that consumers want more clarity about the role AI plays in content production. Baringa found in 2025 that 77% of consumers wanted to know when content had been created wholly or partly by AI, while 31% said they would feel uncomfortable consuming content without knowing whether the technology had been used.
This research is global, rather than specific to Latin America. It therefore does not mean that every consumer in the region expects an AI label on every post, but it does reinforce a broader concern around visibility and authorship.
Transparency alone does not solve everything, and experimental research published in Computers in Human Behavior found that disclosing the use of AI in advertising could, in certain circumstances, reduce brand credibility or purchase intention, particularly when AI-generated content was posted by human influencers. The study examined a specific advertising context, so its findings should not be treated as universal, but they help illustrate an important limitation of labels.
A disclosure can tell the audience that AI was involved. But it does not say whether a claim was properly verified, whether an opinion genuinely belongs to the person whose name appears on it, whether the local context was considered or whether someone took responsibility for the final result.
That is where human involvement becomes most important.
Keeping content human does not mean requiring every sentence to begin in a blank document and be typed manually by an employee. AI can help organise research, produce a first draft, suggest headlines, translate materials or adapt formats. These uses do not automatically undermine credibility.
The decisive point comes afterwards. Who checked the claim? Who decided whether the argument accurately represented the company’s position? Who realised that a particular choice of words worked in Mexico but sounded wrong in Brazil? Who noticed that an image was technically convincing but culturally inappropriate? Who decided that a customer story should not be synthesised because the fact that a real customer had lived that experience was precisely what gave the story value?
Human involvement has less to do with who pressed the keys and more to do with where judgement and responsibility sit.
What this means for brands in Latin America
Sherlock Communications’ own research, Cybersecurity: Data Protection and AI in Latin America, adds another layer to this picture. Based on responses from more than 3,400 people in Argentina, Brazil, Chile, Colombia, Mexico and Peru, the study found that almost half were still cautious about sharing personal information, while 63% believed the growth of AI would increase data fraud.
This matters because audiences do not encounter AI-generated content in isolation. Their perceptions are being shaped at the same time by concerns involving fraud, manipulation, privacy and the use of personal data.
Brands cannot control that broader environment, but they can control how easy it is for audiences to understand their own role within it.
The answer is probably not a blanket rule requiring a disclosure on every caption created with some AI assistance. A more useful principle is proportionality: the more a communication depends on a person’s expertise, authority, identity or lived experience, the clearer the connection to that person should remain.
That means using real experts when expertise is part of the claim, keeping real customers connected to their own stories and giving local teams enough autonomy to challenge content that may be technically correct but culturally wrong. It also means making it clear when realistic synthetic material could reasonably lead someone to believe they are seeing or hearing something real.
Above all, brands should avoid using AI to manufacture the very thing they are asking audiences to trust.
If the value of a message comes from the fact that a founder believes something, a customer experienced something or an expert knows something, the real involvement of those people is not merely an aesthetic choice. It is part of the evidence behind the message.
Consumers in Latin America appear willing to live with AI, but the available data also shows growing uncertainty about what they encounter online and about who is responsible for that content.
AI will continue to give communications teams new ways to create, adapt and scale content. The challenge is ensuring that, as the technology becomes less visible, the judgement and responsibility behind what is published do not disappear with it.