(I have to correct one point: the י here isn't a Nach Nistar but an Eim Mikra, like the ה at the end of תורה, the א at the end of ברא or the ו at the end of בו. A real Nach Nistar י would definitely need to be separated, like in [אדני] ימלך לעולם ועד.)
You're certainly right that Artscroll is wrong from a classical grammar perspective. The Talmud (Berakhot 15b) already listed the cases of הפרדה בין הדבקים separating between the [sounds that] stick [together] in Shema, which are על לבבך, על לבבכם, בכל לבבך, בכל לבבכם, עשב בשדך, ואבדתם מהרה, הכנף פתיל, אתכם מארץ and yours isn't on the list. R. Avraham ben HaRambam was asked about this particular exclusion in a responsum (79:2) and answers that the י in בני isn't really pronounced like a י so there's no issue. (He does seem to indicate that if it was a Chirik instead of a Tzere it would be an issue (eg. בכורי ישראל) but that could just be his funny language used to describe the grammar.)
That said, arguably they pronounced the Tzere vowel as less of a diphthong than is common in modern Ashkenazi Hebrew dialects which are popular among Artscoll's target audience. There is another version (in the Rif, R Amram Gaon, Sefer HaManhig, etc.) of the above cited Talmud which includes the example תזכרו ועשיתם which indicates that diphthongs do indeed count as a sound that needs to be separated. Rabbenu Yonah to the Rif there (8b) prefers our version of the Talmud though, because the final ו in תזכרו is "נח" quiescent. He might mean the end of the diphthong doesn't count as a sound in this regard, but he could just simply be pointing out that it's a mater lectionis and hence a clear error, without commenting on the issue of [non-native] diphthongs.
