Those "Thermal Throttling" values (and some other values under the AMD sensor) is a different case and might have several other reasons. Some of them (especially the EXT) might be triggered by external logic (i.e. the SIO) for whatever reason - either the SIO issues PROCHOT as a result of a valid/invalid thermal evaluation or it receives wrong data. Which of those reasons really happens is difficult to determine as it would require hardware debugging. "Thermal Throttling" might also be triggered by the CPU SMU unit as a result of some firmware erratum or erratic read of temperature data internally in the CPU.
Then there is another case of reading wrong data - running simultaneously with other tools accessing the same resources without synchronization. This has been just recently discovered with NZXT CAM software.
So as you can see, there can be several cases and it's quite difficult to tell whether the erratic value read is a problem of the hardware sensor providing wrong data, or a glitch during reading the value, or some other interaction inside the system.
I don't deny the fact that the erratic values under the ITE sensor might be caused by some recent update in HWiNFO, but I don't think this is something directly related to handling of the ITE sensor. It might be some indirect/rare action/situation that's occurring on a completely different place. But why/where exactly is very difficult to determine due to the complexity of the hardware/software and rare occurring of the erratic values.