Theoretically yes, though the answer to your question may depend on the properties of the two sources.
If, for example, the two sources have different wavelengths, then you need a glass surface with an optical coating that reflects the one wavelength and transmits the other. By the principle of reversibility, if you had one source with the two wavelengths mixed, you would be able to use the same coated surface to split the two wavelengths out, and if you reverse the direction, the paths taken should not change.
From your question though I get the sense that you are talking about two sources with the same wavelength. In that case they need to be coherent and in phase with each other, and you can use a regular 50-50 beamsplitter.
This also follows from the principle of reversibility, though it is not as obvious. If you have one source then you can certainly use a 50-50 beamsplitter to split it into two beams and couple those into optical fibers. So, you must be able to do the same thing in reverse. However as you correctly noted, if you simply aim both beams into the beamsplitter, you'll get two beams out.
The trick is to think of the reversed system slightly differently. In the reversed system, you have two input beams, one of intensity $I$ just like you expected, and a second input beam of zero intensity. These combine in the beamsplitter into two output beams of intensity $I/2$ which are in phase with each other.
Therefore, in the forward case, the two input beams must also be in phase with each other! If they are, then they will destructively interfere to produce a zero-intensity output beam at one port, and constructively interfere to produce the desired output beam of intensity $I$ at the other port.
This is assuming an ideal lossless beam splitter with all beams perfectly aligned, etc. In reality, you will not get perfect transmission (nor would you in the reverse case) but you will get much better than 50%.