I'm under the impression Google Translate uses relative character frequencies to suggest which language.

If you are looking at a page which is mostly hashes, the distribution of the hash values likely affects which language is suggested.