Browser fingerprinting and user behavior tracking are powerful techniques used by most fraud and bot detection solutions. These are implemented via JavaScript snippets running in the user browser. In this presentation, we’ll demystify the signals these snippets collect and describe why these signals are unreliable. Using a realistic threat model, we’ll describe various attacks against defenses relying on these signals. Finally, we'll share war stories of architectural and implementation flaws we found in real world deployments.