I think this question is a little hard to answer definitively. Most programming languages I have used could be used to mislead the developer. It is written in code after all.
I think one of the worst offenders would be JavaScript. JavaScript lets you get away with so many confusing practices. Some examples are things like using a variable without declaring it, scopes not being defined by loops/conditionals - but functions, the strange way threads are handled in JS/browsers, the compatibility between browsers, the name conflicts, it goes on. There are a lot of ways to get hung up.
Now, I think something opposite of that would be a strongly-typed language with an aggressive and ubiquitous IDE like C#. There are many warnings, and "suggestions" that the IDE/compiler provides. I may just be saying that cause I understand C# the best. Java is probably pretty similar to C# in this regard.
Another way of measuring sneakiness would be how side-effects are handled. In that case, a (mostly)purely functional language like haskell might be a good example. Functional langugaes can often be proven correct with significantly less effort than a similar piece of code in a language with a different paradigm.
So a strongly-typed language with a good IDE might be the best way to catch any sneakiness for someone with not much experience in that language. I think if you are an expert in a functional language, it'd be much less likely to get away with any sneakiness there. Maybe the perfect balance is F#!