Transparency about the use of AI is becoming an expectation among customers and citizens

Corporate discussions about artificial intelligence have focused primarily on adoption: which tools to use, which processes to automate, and how much productivity can be gained.

However, as AI becomes embedded in content creation, communications, customer service and decision-making, an equally important question emerges: how should an organization disclose the involvement of these systems?

The results of the 28th edition of Navegantes en la Red, published by the Spanish Association for Media Research, indicate that this concern is already becoming part of users’ expectations.

The statement receiving the highest level of agreement within the survey’s artificial intelligence section was, “It should be possible to identify everything made with AI.” Its average score was 4.17 out of 5, exceeding statements concerning the future potential of AI or concerns about its growing use.

The study was conducted between October and December 2025 and was based on 15,021 valid responses from internet users in Spain. 

This is therefore no longer solely a technical or regulatory issue. Transparency about the use of AI is also becoming a market expectation.

Adoption is moving faster than governance

According to the study, 69.5% of Spanish internet users had used an AI tool during the month before the survey. A further 36.5% had used one the previous day.

In addition to searching for information, respondents use these tools to write text, summarize documents, translate content, and generate images or videos. A significant proportion also report using AI for professional purposes.

These figures reflect a reality that many organizations can already observe internally: AI adoption does not always begin as the result of a corporate decision. Employees frequently introduce these tools themselves to complete specific tasks.

For this reason, prohibiting or ignoring their use is rarely an adequate strategy. Organizations need to understand which tools are being used, what information is being entered, for which purposes, and under what level of supervision.

When adoption precedes corporate policy, risks arise in areas such as confidentiality, data protection, intellectual property, output quality, and consistency of corporate communications.

AI governance begins at precisely this point: transforming fragmented and difficult-to-audit practices into known, assessed, and controlled ways of working.

Utility is not the same as trust

The study also found that 69.7% of AI users had encountered incorrect, unreliable, or false information at least occasionally.

This finding is particularly relevant for businesses. People may continue to use a technology despite recognizing its limitations because its immediate benefits outweigh the effort required to verify every output.

In a corporate environment, however, verification cannot depend exclusively on the individual judgement of each employee.

Whenever content may affect customers, commercial decisions, reputation, regulatory compliance, or third-party rights, oversight must be built into the process. Organizations must define who reviews an output, which sources must be validated, which uses require approval, and what evidence should be retained.

Trust is not created by claiming that an organization uses AI responsibly. It must be demonstrated through procedures, clear accountabilities and appropriate records.

Labeling is only one element of transparency

The demand to identify AI-generated content is often reduced to an apparently simple question: whether to add a label.

In practice, a transparency policy must address a broader set of questions:

• What level of AI involvement should be disclosed?
• Should organizations distinguish between AI-generated, edited, translated, and AI-assisted content?
• Who is responsible for verifying the disclosure?
• How should information about the tools and processes used be retained?
• What happens when suppliers, agencies, or external partners are involved?
• How can the origin of the content subsequently be verified?

The answer will not necessarily be the same in every situation. An internal translation, an advertising image, an automated customer response and a financial report present very different levels of risk.

Effective governance does not mean labelling everything in the same way. It means classifying use cases and applying controls that are proportionate to their potential impact.

Regulation reinforces an expectation that already exists

The European Union’s Artificial Intelligence Act introduces specific transparency obligations for certain systems and types of content.

Among other requirements, it covers the machine-readable marking of certain AI-generated or manipulated outputs. It also introduces disclosure requirements concerning deepfakes and certain AI-generated or manipulated text published to inform the public about matters of public interest.

This does not amount to a universal requirement to label every document in which a generative AI tool has played a role. The precise requirements depend on the type of content, the system involved, and the context in which the content is published.

The transparency provisions under Article 50 will apply from 2 August 2026. The European Commission is developing guidelines and a voluntary code of practice to support their implementation. 

For organizations, the practical conclusion is clear. Waiting until every regulatory interpretation has been finalized does not remove the need to act. The demand for transparency already exists among users, customers, and professionals.

Moving from policies to verifiable mechanisms

A corporate transparency strategy should include at least five elements:

1. An inventory of AI use cases, identifying the tools, processes, owners, and data involved.
2. A risk-based classification system, distinguishing between internal uses, external content, automated decisions and activities with potential legal or reputational consequences.
3. Human oversight criteria, determining which outputs must be reviewed and who retains final accountability.
4. A disclosure and labelling policy, adapted to the type of content, the degree of AI involvement, and the intended audience.
5. Traceability mechanisms, capable of documenting the origin of the content, the transformations applied, and the validations completed throughout the process.

These measures should not be viewed solely as defensive controls. A clear policy reduces uncertainty among employees, supports safer adoption, and enables the organization to use AI more consistently.

Trust will become a competitive advantage

Users do not appear to be rejecting artificial intelligence. They use it, recognize its value, and remain aware of its limitations.

What they are demanding are the conditions required for trust: knowing when AI has been involved, understanding who is accountable for the result, and having access to mechanisms that can verify the origin of content.

The next stage of corporate AI adoption will therefore not depend exclusively on gaining access to more advanced models. It will depend on each organization's ability to integrate AI through transparent processes, clear responsibilities, and verifiable evidence.

Governance should not be understood as a barrier to innovation. It is the infrastructure that allows an organization to innovate without turning every new AI use case into a risk that is difficult to control.