Unproctored Internet Testing (UIT) is a popular form of assessment due to its flexibility and low cost. Results from a UIT may be invalid if there is excessive cheating, but this risk can be acceptably small for certain low-stakes tests. In this paper, we analyze the occurrence of one form of cheating (source-code plagiarism) in a low-stakes UIT for programming aptitude conducted in India with 25,814 test takers. Although we find that the extent of plagiarism is sufficient to threaten the validity of this particular test's results, our analysis suggests feasible ways in which such a UIT can modified to enhance validity by reducing the occurrence of plagiarism.