Great question.
Answer: Radians are actually unitless. In fact, one way to define an angle is to say "it's a (technically nice enough) subset of a circle." We then say that the measure of the angle (which is the kind of thing that we allow as an argument to sine or cosine) is the arclength of the subset divided by the radius of the circle. That's a quotient of two distances, so it's unitless.
The peculiar thing here is that we're very sloppy in naming an angle (in classical geometry: a pair of rays with the same origin; in my definition above, a subset of a circle), and in naming the "measure" of that angle (which is a real number). We tend to use the name "$\theta" for both of these things, which can confound stuff until you get used to it.
Still, one might ask "if radians are unitless, why do we use them at all?" I don't really have a good answer here, except that people also measure angles in degrees (which are also unitless!), and it's nice to have a way to say "I'm converting from degrees to radians", when in fact what we're doing, when we write
$$
\sin x
$$
and $x$ is "an angle in degrees", is that we're really computing
$$
sind(x)
$$
where $sind$ is an entirely distinct function from $\sin$, and can be defined by
$$
sind(x) = \sin (\frac{\pi x}{180}).
$$
You can tell that $sind$ and $\sin$ are different functions, by looking at their values on the real number 180. The value of the first on $180$ is zero; the value of the second is about $-0.8$.
Still, people find the idea of two different functions more baffling than that of two different units, so we pretend radians and degrees are "units" of angle measure.