Definitely a scam.
Sounds like it's playing on the notion of inductive / capacitive loads that have no power factor correction.
It's almost certainly a box containing a switch, resistor and capacitor to provide a voltage to turn on the LED.
I've got a cheap watt meter that derrives it's power that way. It uses the resonance in the RC network to tap off a few volts and a tiny bit of current that powers the meter chip and LCD display. It was cheaper than putting a small transformer in there.
Of course, it could have some big capacitors in there to correct a big inductive power factor error but then it's not connected to the offending motor so when the motor's not running the big capacitors are now creating a capacitive power factor error... Your AC or fridge motor doesn't run all the time as it's governed by a thermostat so when it's not running your Powergard is over correcting for a reactive load that it wasn't sized properly to correct in the first place.
Power factor correctors are used in industry where reactive power from big motors is a problem and can cause power supplies to overload with reactive current or make the utility bill higher. The solution for an automatic power factor corrector that isn't built into the offending bit of kit are the size of a wardrobe and have a bunch of capacitors the size of coke cans with some sensing electronics and a bunch of relays to switch in the right number of "coke cans" to correct the reactive power seen on the line. The key is that you need the right capacitance to correct for the type / size of motor in use. There isn't a one size fits all as the makers of the Powergard seem to think.
http://en.wikipedia.org/wiki/Power_factor
Quoting the Wiki on the random addition of power factor compensators to a power line, it says this:
"In a worst case, reactive elements can interact with the system and with each other to create resonant conditions, resulting in system instability and severe overvoltage fluctuations. As such, reactive elements cannot simply be applied at will, and power factor correction is normally subject to engineering analysis."
Which basically means, don't go bolting big capacitors or inductors on the power grid unless you have to and you've measured what it is you are correcting.
Most modern kit is power factor corrected these days (even computer switch-mode power supplies that used to be the worst offenders for presenting highly capacitive loads). CFL bulbs can still be a problem though but as they consume very low amounts of power anyway the issue of apparent power is largely moot. Some cheap mofified sine wave inverters don't like driving CFLs or other fluorescent tubes because they present a capacitive load to the nasty "sawn-off" sine wave output they have that can cause the mod-sine harmonics to heat up the bulb's electronics. The cheap wattmeter I mentioned started to make a burning smell when plugged into such a mod-sine inverter... That's how I came to take it apart to find out how it worked and why it almost caught fire :D