A material's opacity, refraction, and reflection of EM radiation is highly dependent upon the frequency. Within the microwave frequencies (1GHz-1THz, or 10⁹-10¹² Hz), materials act quite differently event within that range, and we literally see very different effects in the visible spectrum (400THz-800THz, or 4-8 × 10¹⁴ Hz), demonstrated most pointedly by pigments and gel color filters. It's all one phenomenon: electro-magnetic waves.
For one good example, human skin is virtually transparent to NIR, near infrared, light. We see red light therapy on the rise with lots of people experimenting. The light penetrates the skin and enters muscle, causing relief and stimulating healing. And then we have infrared cookers that have high intensity infrared light that, similar to red light therapy, penetrates the upper layers and get absorbed deeper into whatever meat you're cooking, that is until you generate a crust which will absorb more and be less transparent. Your infrared cooker and your red light therapy lamp are using similar frequencies at vastly different intensities.
The same goes for Wi-Fi and milimeter wave 5G versus your microwave oven. The 5G tower would need astronomical power to heat a car to combustion. Don't build your home near it, but a safe distance should reduce the intensity to the point of safety.
Human bodies are quite transparent to the radio frequencies, but absorption does increase with frequency from radio to microwave. At 0.1 GHz, absorption is about 10cm (meaning the EM waves are absorbing at low intensity, and the power density would be quite low).
https://void.cat/d/TnqEGQ1KvaXPeZR23edZjo.webp
Source: https://core.ac.uk/download/pdf/18451914.pdf
I share this not because I do not think caution is necessary with EM, but just to put it into perspective.
### As for 5G used for causing fires.
We should consider the power levels of a 5G tower. Let's say 10kW for easy numbers. That 10kW is not entirely focused, so let's say at just 1m away an object might experience about 5kW/m² max, or about 3-4 times the intensity of the sun at sea level. The power drop-off for EM is inverse-squarely proportional, meaning that power at 1km be 5mW/m², much less intensity than putting an LED bulb against your skin.
Let's say the tower can output 100x power for a few seconds, and you're only 100m away (or a football field for those at home): that would produce a whopping(!) 50W/m², which may hurt but would not be high enough to cause huge temperature spikes. For reference, the inside of a microwave over is on the scale of 10-250kW/m² (1000x), and the sun at sea level is about 1400W/m² (280x).
A 5G tower would have to create an incredibly coherent beam to measurably increase the temperature of anything at distance. As far as I know, beam forming is not nearly effective enough to do that.
