I’d like to warn you in advance about the high level of sarcasm and biting wit in today’s newsletter. But it’s hard for me not to resort to these tools when the result of the work of several hundred people—achieved through exceptionally difficult compromises, work done to make reporting easier for companies—is being butchered at the last minute just so someone can boast, „Look how efficiently we work—immediately after the delegated regulation was published in English, we’re ready with 23 versions in the languages of the Member States.”.
I must inform you right away: Dear Commission, you are not ready at all. What’s more, the rushed publication of unchecked translations creates additional problems. What kind? Here are three examples:
- ESRS S1, point 20(c), in the original requires specifying the number of „temporary employees,” while the Polish translation refers to „temporary workers” in this context. “Temporary employees” is a completely different concept from “employees hired for a fixed term.” The former are classified as non-employees, while the latter are definitely employees. Therefore, if we were to use the Polish translation of the standards, we would be counting a completely different group of people than if we were to use the original English text.
- Annex II contains a definition of „designed recyclability rate,” which provides an example of an event during the product’s use phase that could reduce this rate. In the original text, the example refers to „contamination preventing recycling.” The Polish translation, however, states „recycling preventing contamination,” which reverses the cause-and-effect relationship. After all, the point is not to design a product in such a way that its recycling prevents some form of contamination. The point is that a product may become so contaminated during use that its future recycling will be impossible.
- ESRS 1, AR 14 for paragraph 32 provides a list of information sources on which a company may base its materiality assessment. The original version includes „peer experience” among them. I think we all know who „peers” are; almost every company I know has conducted some form of „peer review” or „peer analysis” during its materiality assessment—that is, a review of what is material at entities comparable to it (companies in the same industry or with a similar business model). Meanwhile, the Polish version of the standards refers to „partners” experience.„ Which partners? Because in the ESRSs, the term ”partner” is reserved for two types of partners: business partners and social partners.
In the Polish translation of the simplified ESRS, I found a total of 30 errors that alter the meaning of legal provisions, as well as an additional 118 errors involving terminological inaccuracies. The latter group is somewhat less serious, but it will add to the workload when reporting. This most often occurs when a term cited in the ESRS in reference to another piece of European legislation was not taken from the Polish version of that legislation in the Polish translation, but was instead translated anew. The definition of „manufacturer of articles” contained in Annex II refers directly to the definition in the REACH Regulation. The Polish version of that regulation defines such a person as „a person (…) who manufactures or assembles an article.” The same definition, as cited in the ESRS, reads in Polish: „a person (…) who manufactures or assembles an article.” I don’t want to prejudge whether the translator of REACH was correct in using the term „assembly.” Perhaps they should have used „installation” instead. But I certainly wouldn’t want to spend hours—or even minutes—in discussions, first internally within the company and later with the auditor, debating the difference between „assembly” and „assembly.”.
After the simplified ESRSs were adopted on July 3, 2026, someone in the Commission’s translation service simply fed their content into an LLM, generated 23 language versions, and then published them. Without any verification or ensuring consistency with other legal provisions. The fact that the result isn’t as disastrous as it was three years ago is probably due solely to advances in the development of artificial intelligence models. The procedure (if there was even any quality control procedure in place at all?) hasn’t changed.
During the 13 months we spent waiting for the standards to be revised in 2023 and 2024, I raised this issue at every meeting with Commission representatives. Ad nauseam. Over the past year, as we were first moving toward developing simplified ESRSat EFRAG and later as the Commission worked on them, I raised this issue again and urged that this time the translation be checked, verified, and of high quality.
I submitted the list of 148 errors I identified to the Commission last Friday. I have been assured that the staff at DG FISMA will do everything in their power to ensure that the Commission’s translation services make the necessary corrections before the text is published in the Official Journal. I hope this will happen; there is a chance it will, since the scrutiny period ends in early September or (if extended) in November. I will certainly keep raising this issue, again ad nauseam and until it is resolved, because I find such a lack of due diligence particularly egregious. The simplified ESRSs are much easier to use than their first version, so the Commission shouldn’t make life difficult for companies by publishing a poor-quality translation into Polish 😊
Polish 
English