I don't see anything wrong with what he's claiming. It should be possible to harvest a couple microwatts of power, which is sufficient to power a low power sensor transmitting occasionally. With the right matching network, and in nearby range of wifi and phones... I seem to remember a paper on this a while back which was able to get into the milliwatt range but only by modifying wifi router firmware and transmitting at over 1W.
Also, it wouldn't interfere with existing networks if it's only a receiver. Absorbing RF doesn't mean the transmitter has to pump more power out to compensate. But apparently it's really easy to become a "mobile technology analyst" and claim things like that. Perhaps I should try that.