It's fairly recent, mostly from people doing a code audit, and possibly emboldened by the upcoming Lemmy 1.0. See e.g. https://lemmy.ml/post/42339089/23619001.

Personally I believe such criticisms will only strengthen the code by polishing some of its roughest spots, increasing transparency by exposing additional configuration options and the like. The PieFed devs are actively soliciting such genuine feedback - see e.g. https://piefed.social/c/piefed_meta/p/1660885/we-re-here-to-answer-any-questions-you-have-about-piefed-s-code.