Other meanings of Sandbox (computer security)
Computer Security
A sandbox is a security mechanism that isolates running programs, restricting their access to system resources and data to limit the potential damage from malicious or untrusted code.1 It provides a controlled environment where software can execute with reduced privileges, often used for testing unverified code or analyzing malware.2
Sandboxing enforces a security policy that limits a program's capabilities, such as file system access, network communication, and system calls. Common implementations include OS-level sandboxes like seccomp on Linux, which restricts syscalls, and AppArmor or SELinux that apply mandatory access control.2 Virtual machines and containers provide stronger isolation by emulating hardware or sharing a kernel with restricted namespaces.3
Sandboxes are used in web browsers to isolate tabs and plugins, in email clients to scan attachments, and in software development for testing untrusted code.1 They are also a key component of zero-day exploit mitigation, as they limit the impact of a vulnerability.4
The concept of sandboxing dates back to the 1970s with the Multics operating system, which introduced ring-based protection to isolate processes. In the 1990s, Java's sandbox model allowed applets to run in a restricted environment, and later, Adobe Flash and PDF readers adopted similar approaches.1 Modern operating systems like iOS and Android sandbox every app by default, and Windows uses Windows Sandbox for isolated desktop sessions.3
Browser sandboxing became critical after the Chrome browser introduced multi-process architecture with sandboxed renderers in 2008, a model now standard in all major browsers.4
Beyond mainstream uses, sandboxes are employed in honeypots to trap attackers, and in fuzzing to safely execute malformed inputs.5 The seccomp-bpf extension allows fine-grained syscall filtering, and Landlock (Linux) provides unprivileged sandboxing for user-space applications.2 In cloud computing, serverless functions run in microVMs like Firecracker to isolate tenants.3
A notable edge case is time-of-check-to-time-of-use (TOCTOU) races, where a sandbox can be bypassed if the resource state changes between checks.6 Also, side-channel attacks like Spectre can leak data across sandbox boundaries, undermining isolation.4
Sandboxes are not foolproof; they can be bypassed through kernel vulnerabilities, logic flaws, or resource exhaustion attacks.6 The principle of least privilege is essential, but misconfigurations often grant excessive permissions. Performance overhead is another concern, especially for I/O-intensive applications.3
Despite these challenges, sandboxing remains a cornerstone of defense-in-depth, complementing antivirus and firewalls.1 Ongoing research focuses on hardware-assisted isolation, such as Intel MPK and ARM TrustZone, to reduce overhead and strengthen guarantees.5
Sandboxing is a dynamic field, with new techniques emerging to address evolving threats.
Help improve the encyclopedia. Reports go straight to the site manager.