I don't know, beyond "get Francophones to work on this"--but part of the point is that, as the article notes, the absence of gendered pronouns (as in Turkish) is neither necessary nor sufficient to prevent sexism.

That said, my guess is that people working on machine learning with French data, or for French-speaking users, would have to work actively to make sure that, for example, the system didn't treat engineering resumes differently depending on whether they used masculine or feminine forms of the relevant words.