This paper explores the limits of static analysis in detecting malicious code, particularly focusing on the effectiveness of advanced semantics-based malware detectors. The authors present a binary obfuscation scheme that relies on opaque constants, which are primitives that allow a constant to be loaded into a register without the static analyzer knowing its value. Using these opaque constants, they develop various obfuscation transformations that obscure control flow, disguise variable access, and interrupt the tracking of register values. The goal is to demonstrate that static analysis alone may not be sufficient to identify malware, as these techniques can be evaded by sophisticated obfuscation methods. The paper includes a detailed description of the obfuscation techniques, including the use of opaque constants and basic block chaining to enhance the robustness of the obfuscation. It also introduces a binary rewriting tool that can apply these transformations to real-world binaries without requiring source code or debug information. Experimental results show that the proposed obfuscation methods successfully evade both commercial virus scanners and advanced static analysis tools, highlighting the need for dynamic analysis techniques to complement static methods. The authors conclude that while static analysis can be improved to handle more advanced obfuscation techniques, there are fundamental limits to what can be decided statically. They emphasize the importance of dynamic analysis to address the challenges posed by sophisticated malware obfuscation.This paper explores the limits of static analysis in detecting malicious code, particularly focusing on the effectiveness of advanced semantics-based malware detectors. The authors present a binary obfuscation scheme that relies on opaque constants, which are primitives that allow a constant to be loaded into a register without the static analyzer knowing its value. Using these opaque constants, they develop various obfuscation transformations that obscure control flow, disguise variable access, and interrupt the tracking of register values. The goal is to demonstrate that static analysis alone may not be sufficient to identify malware, as these techniques can be evaded by sophisticated obfuscation methods. The paper includes a detailed description of the obfuscation techniques, including the use of opaque constants and basic block chaining to enhance the robustness of the obfuscation. It also introduces a binary rewriting tool that can apply these transformations to real-world binaries without requiring source code or debug information. Experimental results show that the proposed obfuscation methods successfully evade both commercial virus scanners and advanced static analysis tools, highlighting the need for dynamic analysis techniques to complement static methods. The authors conclude that while static analysis can be improved to handle more advanced obfuscation techniques, there are fundamental limits to what can be decided statically. They emphasize the importance of dynamic analysis to address the challenges posed by sophisticated malware obfuscation.