I know that in multiparty computation we may consider semi-honest or malicious adversaries. I can find papers that consider malicious adversaries, but the semi-honest term is only mentioned in MPC papers. Are those adversaries particular for MPC, or used also in other technologies, such as homomorphic encryption, differential privacy etc.
Yes, malicious adversaries are not specific to MPC. Consider the security of encryption schemes. You can view chosen-plaintext attack (CPA) as semi-honest security and chosen-ciphertext attack (CCA) as malicious security. The former does not allow the adversary to modify the ciphertext, but the latter does and the adversary may learn extra information from the decryption oracle. You can say something similar for key-exchange vs authenticated key-exchange since plain key-exchange is not secure when you have a malicious adversary which could do a man-in-the-middle attack, that's why we need authenticated key-exchange. In fact, semi-honest/malicious adversary occurs in any cryptographic protocols like secure messaging, TLS and much more.