As evidence, the lawsuit cites unnamed “courageous whistleblowers” who allege that WhatsApp and Meta employees can request to view a user’s messages through a simple process, thus bypassing the app’s end-to-end encryption. “A worker need only send a ‘task’ (i.e., request via Meta’s internal system) to a Meta engineer with an explanation that they need access to WhatsApp messages for their job,” the lawsuit claims. “The Meta engineering team will then grant access – often without any scrutiny at all – and the worker’s workstation will then have a new window or widget available that can pull up any WhatsApp user’s messages based on the user’s User ID number, which is unique to a user but identical across all Meta products.”
“Once the Meta worker has this access, they can read users’ messages by opening the widget; no separate decryption step is required,” the 51-page complaint adds. “The WhatsApp messages appear in widgets commingled with widgets containing messages from unencrypted sources. Messages appear almost as soon as they are communicated – essentially, in real-time. Moreover, access is unlimited in temporal scope, with Meta workers able to access messages from the time users first activated their accounts, including those messages users believe they have deleted.” The lawsuit does not provide any technical details to back up the rather sensational claims.
Even in an “insecure” app without air-gapped systems or manual encryption, creating a backdoor to access plaintext messages is still very difficult if the app is well audited, open source, and encrypts messages with the recipient’s public key or a symmetric key before sending ciphertext to a third-party server.
If you trust the client-side implementation and the mathematics behind the symmetric and asymmetric algorithms, messages remains secure even if the centralized server is compromised. The client-side implementation can be verified by inspecting the source code if the app is open source and the device is trusted (for example, there is no ring-zero vulnerability).
The key exchange itself remains somewhat vulnerable if there is no other secure channel to verify that the correct public keys were exchanged. However, once the public keys have been correctly exchanged, the communication is secure.
Well audited is key, this attack likely works by doing something like adding Meta to the list of trusted devices, then hiding itself from the list (either because of code in the client or because it the meta device is only added for a moment), so the backdoor wouldn’t be send_all_messages_to_hq(), it would be in the code to list trusted devices, either explicitly hiding some devices or some sort of refresh timer that’s known so you can avoid being there when the UI is updated).
Or it works through the some other mechanism that still preserves E2E encryption.
I assumed that not only the entire app but also the entire client device had been audited. This was a client-side attack, not Meta momentarily adding itself to the trusted-device list. I’m confident it was a client-side attack because it would be impossible to hide even a momentary change in keys from the client without modifying the client app to conceal such a change.
Does WhatsApp make it visible when you add a new trusted device? Does Signal?
But yeah Meta have full control of the client and it isn’t audited so they could do it a lot of ways.
Yes, it is visible when a new trusted device is added. The QR code you scan to link a device contains a one-time public key for that device (ECC is used partly to fit the public key more easily into a QR code). Signal on the phone then sends a lot of information, including the identity keys, to the new device. The new device uses these identity keys to communicate. Note that the transfer of identity keys is fully encrypted, with encryption and decryption taking place on the clients. This can, of course, be bypassed if someone you’re talking to has their security key compromised, but the same risk exists if the recipient takes a screenshot or photographs their device’s screen.
Edit: The security key refers to the one-time key pair generated to initiate the transfer of identity keys and chat history. It can be compromised if someone accidentally scans a QR code and transfers their identity keys to an untrusted device.