Padding oracle attacks exploit information revealed during the decryption process when padding is checked. In block cipher modes that require padding, the system verifies whether the padding bytes are correct after decryption. If an attacker can alter ciphertext and observe the server’s response—whether it flags a padding error or proceeds normally—that response acts as an oracle about the padding validity. By carefully manipulating the ciphertext and interpreting these padding-valid vs padding-invalid signals, the attacker can recover plaintext byte by byte without the key. This specific leakage, tied to how padding is validated, is what makes it a padding oracle vulnerability. Other concepts don’t describe this exact leakage. Side-channel attacks rely on indirect clues like timing or power usage rather than padding correctness; replay attacks resubmit captured messages; and man-in-the-middle attacks focus on intercepting and possibly altering traffic but don’t inherently depend on padding validation leaking plaintext.