Violating assumptions with fuzzing

Fuzzing is a highly automated testing technique that covers numerous boundary cases using invalid data (from files, network protocols, API calls, and other targets) as application input to better ensure the absence of exploitable vulnerabilities. Fuzzing lets developers or quality assurance (QA) teams test large numbers of boundary cases when doing so with techniques such as functional testing would be cost prohibitive. Comprehensive negative test cases - those that verify that a product does not do something it shouldn't do, rather than that it does something it is supposed to (positive test cases) - are difficult to construct because the number of possible permutations is astronomical. Yet, fuzzing covers a significant portion of negative test cases without forcing the tester to deal with each specific test case for a given boundary condition.