Post an example of a position where this happens.
From your description, my prediction would be that the position was already very one-sided and the computer attempted to give up material to delay checkmate, or something like that. Otherwise, giving up material is usually a bad move.
Why is it that the engine that generates the analysis post game is always listing moves that would have resulted in lost material as mistakes. Sometimes it seems to completely hallucinate. If I am correct these results factor into your accuracy percentage which means you often have results WAY off from what they should be.
I think I am right about this. Please correct me if I am wrong because I am tired of doing reviews which have good moves listed as mistakes because I failed to take a pawn that would have cost me important material. I mean looking forward in the game I am just not seeing where this makes sense.