Yes, this is bad. I have seen discussions in which attackers quickly asked about device console logs. Using the strings from those logs, they can quickly search the executable binary for similar strings, like the format string used for part of the log entry. Example:
NSLog(@"user entered password '%@'", passwordString)
Finding the format string, they can quickly find the part of code that actually uses that string, and thus they have found the part of code that processes passwords. They can then analyze the relevant code to bypass the password altogether or do whatever they want.
Even if the device does not hold a lot of console data for long, all an attacker might need to do is connect a stolen device to a computer, start the app and enter a random password. If the app logs the password (correct or incorrect), then the needed strings are still captured and the attack begins.
Of course, if the log line is still in the device console data, an attacker also gets the actual password for an actual user. And if that user reuses user names and passwords (as many people do) the attacker may now have the credentials to a user's other accounts on the internet, which may be more sensitive.
Many programming discussions advise against using NSLog directly and recommend using a macro like DEBUGLOG which will compile out of release builds. If that is used for debugging, then the app will not leave such clues for an attacker to use.