Semantic security
The requirement that a ciphertext reveals nothing about its plaintext, not merely that the key is hard to guess.
Shafi Goldwasser and Silvio Micali formalised the notion in 1982, in their work on probabilistic encryption. A scheme is semantically secure if an adversary learns nothing about the plaintext from the ciphertext beyond what it could have guessed anyway, even when it can choose plaintexts and see their encryptions. A direct consequence is that deterministic encryption can never meet the standard, since the same plaintext always produces the same ciphertext.
ECB fails for exactly that reason, as the penguin picture shows. Random IVs and nonces are the usual way to meet the requirement in modern modes.