Only you can decide if your robots can have their minds read. The preponderance of fiction that I have encountered would suggest that artificial intelligence is immune to mind reading. However, mind reading versus robots is completely up to you. Your story, your world. That being said...
Password Protection is Likely Irrelevant to Telepathy
To the extent that your robot thinks, and that your telepaths can perceive and decipher the thoughts of others, retinal scans and passwords are not going to add security. A telepath isn't plugging in through some kind of access port, nor yet attempting to get the robot to willingly give up its thoughts. There is no front door or back door to a human mind, so you would assume that a telepath is skipping straight to the MIND. No keys or locks or encryption would be meaningful if the telepath is using the robot's own understanding of itself to understand the robot.
Biology May Be Too Different
You could argue, on the other hand, that whatever kind of thinking the robot does, the mechanism is too foreign for understanding by telepathy. Maybe the telepath links to the soul, and the robot has no soul. Maybe telepathy interprets each individual's unique neurological structure, but microchips, or whatever the robot has, just don't make sense by the interpretive logic of psychic abilities.
However, if the mind of the robot cannot be read at all, your psychics are sure to at least notice that something is very, very wrong with this individual.
It would be interesting if the telepaths could get at whatever the robot explicitly thinks about, but could not guess at what unretrieved information was buried in its memory banks. Even if they could perceive the robot's thinking, it might be organized in a way that is very different from the loose, associative way people tend to think about things.