The Implied Association Tests are not reproducible, meaning when the same person takes the same test at different times, they get very different results. I don't know if they can be "gamed" though I imagine they could be, as the test measures response delay, supposedly to show that someone is consciously trying to compensate for their unconscious bias. So, I think you could falsify your result by intentionally slowing down or speeding up your responses.