Err no, many experts have stated the system is very easy to trick. And of course, the scanning list will be injected with other things it needs to find, without the knowledge of Apple and others. At which point, a person at Apple will be able to see the image, not a hash, the actual image.
I think those experts will be proven wrong. It's pretty hard to be sure when Apple hasn't released the code or the gritty details of it. It's impossible to know for sure how good NeuralHash is without testing it. The only one who has is Apple which is why I trust them more until there are actual tests outside Apple.
Also, it's not easy to insert new hashes into an iPhone without Apple's knowledge. There are many ways to check if the hash table has been tampered with.
In addition, Apple can overwrite the hash table with a new version every time you update the OS.