The page says that the 10O resistor current limits the circuit to 50mA, but I don't think that's correct: 5V/10O = 0.5A = 500mA.
edit: never mind, he said charging current, so the battery enters into the equation too.
I have other concerns about the safety of this circuit. Most notably, I think it may create an voltage spike when switching on/off. I'd need to look up some battery specifications to be more sure, but it appears that nothing prevents the voltage from going above 3.3 if the battery's resistance becomes too high (the battery acts like a variable resistor, so the circuit would act like a variable voltage divider), so it seems that the voltage could spike between the time that the battery controller switches over to the battery circuit and the time that the 5V rail discharges.
On the other hand, this might not ever happen in practice, or the spike might be confined to such a short timeframe that the circuit isn't materially affected by it. But I see nothing that stops the spike from happening in the first place...
edit: it occurs to me that the overvoltage condition should occur while power is on as well, and he's been using it for months. So disregard my rantings I guess. :looney