First, teach them how to do informed estimates for the probabilities of these negative factors coming into play.
Then, ask them to point out the probability-estimate below which they won't even bother to bring something up.
Finally, explain that your bar for this is higher than theirs, and request that they only bring things up if they meet your higher bar; for things that meet their bar but not yours, you'll just be discarding what they say anyway, so they'll just be wasting their breath.
As someone who's been in this business for 30 years I've come to recognize that "unlikely failures" is just a synonym for "pretty sure it's going to happen at some point".
Also, there are way too many programmers around that are the exact opposite, and don't see any of the obvious ways in which a design will fail, so I tend to be kind of protective of those that do.
I like working with them and having at least one on the team, because they're a great addition to the "naive optimist" or the "carefree hacker-fixer" who always see the solution around the corner - in a couple of minutes of course. ;)
And: There's a lot of tech scenarios where being overly careful and actually seeing the impending technical doom in every corner is an advantage.
Similar to the pedantry you get by the human robot, Cassandras have their place - and good management and coworkers know exactly where and how to take them.
I remember the first time I encountered this, and thinking the guy had to be on crack.
His quote: "A changed comment in code could possibly cause a problem, if the file were to become corrupted as its uploaded"
The tension probably arises from disagreement over what is critical.
Healthy skepticism is beneficial, but I've seen teams paralyzed because one member has perfected the technique of raising unanswerable or irrelevant objections, often seemingly to present the appearance of being the smartest guy in the room.