This Security Researcher Is Thoroughly Incompetent
The "security researcher" thinks that the password must exist somewhere on the device, but that (a) is not true for any secure implementation, and (b) is not true for the Samsung devices that have been validated by NIAP (S4, S5, etc). The "researcher" first configured the device to be completely insecure by allowing PINs instead of passwords, and allowing password hints, neither of which is allowable under a secure configuration.
The "researcher" is demonstrated to be even more incompetent by the fact that s/he did not even bother to check public domain documentation on the Samsung / KNOX security, which would have led him / her to realize that unlocking the user data on the device, once it has been turned off, would require physically dismantling the device and probably modifying some silicon.
It is unfortunate that the tech media does not have the intellectual wherewithal to properly investigate such claims and instead trumpets them without understanding. Instead, attention should be properly focused on the fact that certain smartphone manufacturers rely entirely on security through obsecurity, using legal stratagems to deny *actual* security researchers the right to investigate their devices. Google "Charlie Miller" to see an example of this.