Ask ten translators and you will get ten answers. Ask ten clients and you will mostly get one: yes. A case for disclosure as a professional standard, and the best arguments against it.
THE SHORT VERSION
- Our view: yes, and not as a confession but as a standard.
- Clients decide the risk, but only if they know what is happening to their text.
- Disclosure protects translators too, by defining who is responsible when something goes wrong.
- The real question is not “AI or human”. It is which tools, at which step, with which human review.
Ask ten translators whether they should tell clients when they use AI, and you will get ten answers. Some will say it is none of the client’s business, the same as any other tool. Some will say it is an ethical duty. Some will say it depends on the tool, the text and the client. Ask ten clients the same question and you will mostly get one answer: yes.
Our view: yes. And not as a confession, but as a standard.
Why disclosure matters
First, clients decide the risk. A client who knows machine output is involved in a job can ask for more review on a contract and less on an internal memo. They can choose a different workflow for a medical leaflet than for a staff newsletter. A client who does not know has no way to make that choice. The decision is taken for them, silently, by the supplier.
Second, it protects the translator too. If a tool leaks confidential text, or introduces an error into a sensitive document, a clear agreement defines who is responsible and what the client accepted in advance. Without one, the conversation after an incident starts from distrust.
“The question isn’t “AI or human.” It’s “which tools, at which step, with which human review.”
Third, disclosure ends the fake binary. Almost nobody works at either extreme. Most professionals use some combination of reference tools, memory systems, machine suggestions and their own judgment. A flat claim of “100% human” invites suspicion, and a flat claim of “AI-powered” says almost nothing. Describing the workflow honestly, step by step, is more useful than either slogan.
The best argument against
The strongest counter-argument deserves a fair hearing. Translators have used computer-assisted translation tools, spell-checkers, dictionaries and search engines for decades without announcing each one. Nobody expects a translator to write “I used a dictionary” on an invoice. Where, exactly, does a tool become something the client should be told about?
There is also a practical worry. Some translators fear that disclosure becomes a reason to cut rates, or that clients will misunderstand how much human judgment remains in an AI-assisted workflow. Those concerns are real, and they are a reason to explain the process carefully, not a reason to hide it.
Where we would draw the line
Our line is this. Disclose when the tool generates the text that goes to the client. Disclose when the client’s content is uploaded to a third-party system, because that is a confidentiality question as much as a quality one. For tools that only assist the human, such as dictionaries, spell-checkers, terminology lookups and translation memory, a standing description of the workflow is enough.
What disclosure can look like in practice
- A short workflow statement in the quote or contract: which steps use machine output and which are done by a person.
- A clear statement about what happens to the client’s files, and which systems they pass through.
- A choice, where it matters: the client can ask for a human-first workflow for sensitive content.
- Named human responsibility: who reviews, and who answers if something is wrong.
None of this requires a long document. A few honest sentences, written once and updated when the process changes, can do most of the work.
We could be wrong
We could be wrong about where the line sits. Technology is changing quickly, and a tool that is peripheral today may be central next year. We would rather argue about the line in public than leave it undefined.
So we would like to hear from you. Where would you draw it? Translators: would disclosure change how you work? Clients: would it change how much you trust the result?
Comment below. We answer every comment.
Language Shift by Transinity: AI, localization, careers and culture from inside the language industry. Follow Transinity on LinkedIn and join the conversation.