Your Voice AI Has One Week Left to Stay Silent About Being AI
The most dangerous thing your AI assistant may say is not something incorrect. It may be what it fails to say at all: “I am an AI.”
From 2 August 2026, transparency obligations under Article 50 of the EU AI Act will apply. Businesses using certain AI systems to communicate with customers must ensure that people understand when they are interacting with a machine. The rules also introduce marking and disclosure requirements for specific types of AI-generated or manipulated content.
For companies, this is a question of trust.
In her latest article for Fast Forward Magazine, Marianna Konina, Founder and CEO of Reputation City, explores this approaching deadline together with Kyriaki Parmakki, co-founder of Essere.ai. The discussion examines what the new requirements mean in practice, which businesses may be affected, and why transparency should become part of a company’s reputation strategy.
The AI Act Was Delayed – But Not This Part
Recent headlines about delays to parts of the EU AI Act may have created a dangerous sense of relief. While the implementation timetable for certain high-risk AI systems has changed, the transparency obligations under Article 50 are still scheduled to apply from 2 August 2026.
These rules are intended to help people recognize when they are interacting with an AI system or encountering certain AI-generated content.
This distinction matters because Article 50 is relevant to technologies that are already widely used in ordinary business operations. These can include customer service chatbots, voice assistants, automated messaging tools, and other AI systems that communicate directly with people.
The problem is that many companies may have heard only one message: the AI Act has been delayed. As a result, they may not realize that the transparency deadline affecting their everyday tools remains in place.
An Invisible Deadline Can Still Create Real Risk
Nothing dramatic is likely to happen the morning after the rules begin to apply. There may be no immediate inspection, warning or technical interruption.
That is precisely what makes the situation risky.
An undisclosed AI assistant may continue answering questions, processing inquiries and speaking to customers as usual. However, each interaction could become another example of the company failing to meet its transparency obligations.
A complaint from a customer, a report from a competitor or closer attention from a regulator may suddenly expose a problem that has existed for months. By that point, the issue is no longer simply a missing sentence in a chatbot greeting. It may become evidence that the business did not take transparency seriously.
The potential penalties for breaches of certain AI Act obligations can reach €15 million or 3% of worldwide annual turnover for companies, with proportionality provisions applying to smaller businesses. Yet the financial penalty is only one part of the risk.
For many organizations, the reputational consequences may be more difficult to contain.
Which Businesses Should Pay Attention?
The transparency rules are not relevant only to technology companies.
Hotels may use automated assistants to manage reservations. Clinics may have bots that arrange appointments. Real estate agencies may rely on AI to communicate with potential buyers. Financial companies, forex businesses, online shops, and professional service providers may all use automated tools to answer customer questions or qualify leads.
A small company with one AI receptionist should not assume that its size automatically exempts it from its obligations.
The practical question every business should ask is simple: Can a customer interact with an automated system without clearly understanding that it is AI?
If the answer is yes, or if nobody in the company knows the answer, then the organization needs to examine that interaction.
Businesses should also review how they create and publish synthetic content. Article 50 contains separate requirements concerning machine-readable marking by providers and disclosure by deployers for categories such as deepfakes and certain AI-generated text on matters of public interest. The exact obligation depends on the system, content and role of the organization, so companies should confirm their position with a qualified legal adviser.
Transparency Cannot Be Hidden in the Terms and Conditions
A disclosure is not effective if customers are unlikely to see it.
Placing a sentence in a privacy policy or terms-of-service page may not be enough. The person should be informed clearly, and no later than the moment of their first interaction with the AI system.
For a chatbot, this could mean introducing it as an AI assistant in the first message. A voice agent could make the same disclosure at the beginning of a call. The information should be understandable and presented in the language used for the interaction.
Parmakki explains that Essere.ai has incorporated this principle directly into its products. Its AI agent identifies itself from the beginning and is designed not to present itself as human, even when asked directly. The company also maintains relevant records and uses EU infrastructure and data residency as part of its broader compliance approach.
This shows that transparency does not have to damage the customer experience. When included during the design process, it becomes a natural part of the conversation.
Compliance May Begin With One Conversation
For many businesses, the first compliance check does not require a major audit or expensive development project.
A company can begin by opening its own website chatbot, WhatsApp assistant, or voice system and interacting with it as a customer would. It should ask directly: “Am I talking to a human?”
The answer can reveal a great deal. Does the system identify itself honestly? Does it avoid the question? Is the disclosure present in its opening message? Does it work in every language offered to customers?
The business can then document what the system does, who is responsible for it, and how its transparency is checked. Employees who supervise the tool should also understand the change and know how to respond if a customer asks about it.
For a company using only one relatively simple chatbot, the required changes may be straightforward. That makes continued inaction increasingly difficult to justify.
Do Not Switch AI Off. Make It Honest
Fear of regulation can push businesses towards the wrong conclusion.
The purpose of the transparency obligation is to prevent AI from quietly presenting itself as human when people have a right to understand the nature of the interaction.
Switching off every automated system would mean losing the speed, availability and efficiency AI can offer. The more sensible response is to map where AI touches the customer journey and introduce clear safeguards.
This process can produce a wider business benefit. Many companies have adopted AI tools without maintaining a complete record of where they are used, what information they process or who controls them. Preparing for Article 50 forces an organization to answer those questions.
In this sense, compliance becomes part of better AI governance.
When Trust Is the Product, Disclosure Is Part of the Service
The reputational risk becomes especially serious in industries where customers share personal, sensitive, or financial information.
Imagine a patient discussing an appointment with what appears to be a member of a clinic’s staff, only to discover later that the conversation was handled by an undisclosed AI system. The immediate question may be: What else has the organization failed to tell me?
That doubt can spread beyond one interaction.
In a relatively small and interconnected market such as Cyprus, negative experiences travel quickly. A regulatory breach can become a news story, a social media discussion or a reason for customers to reconsider a company’s credibility.
At Reputation City, we believe that responsible AI communication is becoming a fundamental part of reputation management. Customers do not necessarily reject automation. What damages trust is the feeling that automation was hidden from them.
A company that openly explains how it uses AI demonstrates confidence, responsibility, and respect for its audience. Transparency can therefore become more than a legal requirement – it can become a competitive advantage.
The Time to Speak Clearly Is Now
The most practical step a business can take today is to test every AI-supported customer interaction from the customer’s perspective.
Find the chatbots, voice assistants, and automated messaging systems. Check what they say in their opening message. Ask whether they identify themselves truthfully. Review the company’s synthetic-content workflows and establish who is accountable for every AI touchpoint.
The businesses best prepared for the next stage of AI regulation will not be those that most successfully hide their technology. They will be those that use it openly, responsibly and without asking customers to guess whether there is a human on the other side.
The information in this article does not constitute legal advice. AI regulation continues to evolve, and businesses should confirm their specific obligations with a qualified legal adviser.
Read the full original article by Marianna Konina in Fast Forward Magazine.
To learn how automation can free specialists from routine work without replacing human ideas, read our related article: AI Does Not Replace Creativity. It Gives It Time to Work.