TranQuality Labels

Diesen Artikel auf Deutsch lesen

In the AI era, many companies rely on automated translation to publish content in multiple languages. While this technology offers speed and convenience, it is widely acknowledged that machine translation and AI are far from perfect and can produce errors of varying severity. Because modern AI translation systems often generate texts that feel fluent and polished, many assume that automated translation has become a solved issue. In reality, this apparent maturity can be misleading.

Today, browsers, apps and platforms frequently translate content automatically, often without notice or consent of the readers. Yet publishers who rely solely on AI cannot guarantee the accuracy of the translated information, especially when they do not master the target language. Fluency is not accuracy. And when information travels across languages without proper verification, the consequences can be serious.

This growing need for transparency is now reflected in legislation: Article 50 of Regulation (EU) 2024/1689 (the AI Act) requires AI-generated or AI-manipulated content without human editorial oversight to be appropriately disclosed in certain situations.

TranQuality Labels build on the same principle by making the verification status of translated content immediately visible to readers.

The Initiative

TranQuality Labels are a collective initiative of the TranQuality Council, led by a group of translation quality specialists surrounding Alan K. Melby, Gabriela Escarrá, and Stephen Lank. The initiative builds on decades of normative work on translation quality and addresses a growing need in the AI era: clearly distinguishing verified translations from unverified ones.

The Aim: Protecting the Information Consumers

Anyone reading information online can be considered an information consumer. When readers are neither informed that a text is a translation nor can verify how it was produced, they may be misled by the apparent fluency of AI-generated content.

Translation errors are not merely stylistic issues. They can cause real harm, for example, in the context of mechanical systems, medical interventions or legal procedures.

Furthermore, AI systems are known to produce hallucinations and reproduce biases that only a competent human reviewer can detect and correct.

Publishers solely relying on AI cannot guarantee the accuracy of the texts they release in foreign languages if they don’t master them. The AI Safety Report 2025 lists high-risk security issues, citing observed and documented incidents with their level of impact.

Professional language service providers, by contrast, stand behind the quality of their work. They are accountable for what they deliver. With raw AI or machine-translation outputs, such accountability simply does not exist.

For that reason, informing readers about the origin and verification status of translated content is essential. Transparent labelling enables readers to assess the reliability of what they read.

Launched in 2022, the TranQuality initiative anticipated today's regulatory focus on transparency. It complements the disclosure obligations introduced by Article 50 of the EU AI Act with an additional quality label that distinguishes professionally verified translations from unverified ones.

Grades of Translation Quality Labels

To address this need, the Council proposes a simple two-label system:

• Professionally Verified Translation
• Unverified Translation

Professionally Verified Translation (PVT)

The label Professionally Verified Translation indicates that

• a qualified bilingual professional has verified the translation according to frame specifications,
• accountability is traceable to a named human verifier/reviewer or language service provider.

“Qualified” refers to demonstrated competence in the relevant languages and domain, as well as the ability to verify compliance with specifications such as accuracy, completeness, terminology, style and regulatory constraints.

The PVT label is particularly relevant for high-impact scenarios, where errors may have financial, legal, safety or reputational consequences.

Rather than replacing existing standards (such as ISO translation standards), the PVT label complements them by making translation quality visible and tangible to end-users.

Unverified Translation (UVT)

The label Unverified Translation indicates that no qualified bilingual verification has been performed.

This includes:

• raw AI or machine translation output
• translations produced by non-qualified individuals.

Such translations may still be acceptable in low-impact contexts, provided that risks are minimal and appropriate safeguards are in place. However, readers should be clearly informed that the content has not been professionally verified.

Approaches to notify the information consumers

Labelling may take several forms, for example:

• digital watermarks,
• cryptographic metadata,
• a visible notice in a disclaimer, imprint or legal statement on a website or document.

The provided transparency informs readers how the information they read has been produced.

TranQuality by ji

Over the years, I have written extensively about the risks associated with excessive reliance on AI and machine translation. In my daily work as a translator and reviewer, I regularly encounter AI-generated texts that appear convincing at first glance but contain subtle inaccuracies, terminology errors or misleading formulations.

These are often the kinds of issues that only careful human review can detect — the devil in the details.

During more than twenty years as a freelance translator, my clients have relied on my attention to detail, thorough research and precise terminology choices. This trust has allowed me to contribute to demanding projects where accuracy and clarity truly matter.

For this reason, I now offer my clients the possibility to explicitly signal the human-verified quality of the content I translate or review.

Upon request, clients may add my own TranQuality logo to the documents or websites I have translated or reviewed. This mark informs readers that the content has undergone professional verification and meets defined quality standards.

This could look like the following on a website:

A simple step toward trust

TranQuality Labels make translation quality visible.

With the entry into force of the EU AI Act (Regulation (EU) 2024/1689), transparency is becoming an increasingly important aspect of AI-generated content. While TranQuality Labels are an independent initiative, they pursue the same objective: helping readers understand how multilingual content has been produced and whether it has been professionally verified.

Beyond strengthening trust, TranQuality Labels can also help organisations demonstrate their commitment to AI transparency: By clearly indicating whether content is Professionally Verified (PVT) or Unverified (UVT), publishers enable readers to assess potential risks, reduce misunderstandings and strengthen confidence in their brand.

Entrust me with the translation or professional verification of your critical content and display my TranQuality mark on your publications, so your readers know exactly what level of reliability they can expect.

Because in multilingual communication, trust begins with verified words.

Let’s put transparency to work for you — and for all your readers.

Visit my website