They put a “technical device” on some servers to be able to access the messages. Assuming it really was end-to-end encryption they must either have had access to the keys (through a cracked device?) so that they could decrypt the messages or the crypto implementation had some sort of weakness they could exploit such as being open to padding oracle or timing attacks or mitm attacks.
Unfortunately I doubt they will release any details about how they cracked it. It would be interesting to know for sure.