The th sound is a lisp.
Place the tip your tongue between your teeth but make sure the top part has a bit of space to let air flow (like 2 mm). Breath out. Some english words have a vocalized th sound while some don't. That means instead of breathing you put in your voice.
On an unrelated note, if you vocalize a T, it makes a D sound. Vocalizing an S makes it sound like Z. Just thought I'd give you this example so you have a better idea of what I meant by "vocalizing".
I blame the Text-To-Speech they're using. Some of the other courses have bad TTS, but the Italian one is the worst, in my experience.
I wish they'd invest in getting a better Italian voice. The language is much in demand, and the course is otherwise great, but I think pretty much everyone agrees that it's necessary to improve its listening component.
By the way, the new Irish course (currently still in Beta - Sep. 1, 2014), has real recordings of a human native speaker, which is very exciting.