A confidentiality leak is an unauthorized disclosure of data. In software security, it often means an application reveals information it should have kept private, such as secrets, session data, file contents, or fragments of process memory. The leak may be direct, like printing protected data, or indirect, such as reading past a buffer and exposing nearby memory.
Confidentiality leaks matter because even small disclosures can help attackers map a target, recover credentials, or bypass later defenses. In real attacks, malformed files, crafted network messages, or parser bugs can trigger the leak when the program processes untrusted input. Defenders look for strict input validation, bounds checks, memory-safe coding, and sandboxing, because reducing what an attacker can read is just as important as preventing crashes or code execution.