Tor Browser only protects your IP address.

Emails received from outside senders are only end to end encrypted if the sender is using OpenPGP or S/MIME. Otherwise, Proton receives them in plain text (the TLS encryption is terminated at their SMTP server). They promise that they don’t look at them before encrypting them for storage, but you have to trust that promise.

Injecting malicious code means either XSS or if their build pipeline gets hacked. These companies release builds through a pipeline (usually download source -> download dependencies -> build from source -> package -> sign -> notarize (for Apple) -> release), and anywhere along that pipeline can be vulnerable. They might update a dependency that got hacked and now they’re hacked too. One of their build servers might get hacked and now they’ve released a malicious build. You’re trusting them to verify not only their code and their build servers, but also every dependency update. That’s potentially millions of lines of code per year to vet. It’s probably fine, but you’re still trusting them.

As for whether an attack is their fault, it really doesn’t matter. The end result is your leaked data. They could do everything they possibly can to protect you, but they could still get hacked. You are trusting them when you use their service. I believe they’re trustworthy, which is why I’ve been using their service for years.

A note about me: I know all of this because I have worked in big tech for ~11 years (Facebook, Google, then LinkedIn), I wrote an end to end encrypted messenger (called Tunnelgram, now discontinued), and I wrote my own email service over the past two years (called Port87).