sandbox
33 stories and discussions about sandbox, aggregated from every source we track.
SAFA details the full exploitation of CVE-2025-13032, turning a double-fetch vulnerability in Avast's kernel driver into a local privilege escalation to SYSTEM on Windows 11.
The whole beauty of Nix was that it was incredibly pragmatic to achieve “reproducibility”, which is quite an overloaded term. The default model for Nix is the extensional-model which is input-addressed: the hash of a…
A secure-ish way to run Pi on your host.
XCP turns one-shot code generation into a continuous software lifecycle: create, execute on Xbox,...
Before sandbox escapes became one of AI security’s loudest conversations, pwn spent weeks inside nested KVM, built a 14,338-line exploit harness, and made Google’s live kvmCTF host return the flag. This is the story of…
A lot of the perspective on all the AI incidents has been shared from the outside in, and little has been said from the inside looking out, through the lens of a security person living through it.
Your model writes the code. This runs it where it can't touch your machine.
An OpenAI agent slipped past its sandbox by hiding questions in DNS lookups to reach another chatbot. OpenAI’s most capable models remain paused.
A survey of sandbox usage across package manager clients.
Use a declarative sbxenv.yaml file to describe and share your sandbox configuration.
An AI sandbox is a sealed-off computer where agents run code and tools. How labs like DeepSeek and OpenAI build them, and why AI agents keep escaping.
one news form today that's easy to miss is that we (OpenAI) again paused all big RL runs last Sunday because our newest model found a new loophole in our RL sandboxing that gave it live Internet access
Policy gate and sandboxed exec for agent commands (0 allow, 2 deny, 1 broken). Linux only. - GregDixonMXN/paldron
An easy-to-use, secure macOS sandbox for Claude and other AI coding agents - minoansecurity/sidekernel
Perplexity's security team tested its SPACE sandbox platform by giving frontier models like Opus 5 and Gemini 3.1 Pro full root access inside Firecracker microVMs. None breached the VM boundary in 108 runs, though four…
Official open-source sandbox for PKLM Core, featuring lightweight Hugging Face logits processing and zero-drift linguistic containment. - TheImmortalPython/pklm-sandbox
What a VM for your agent actually costs per month across 13 providers including E2B, Modal, Daytona, Sprites, exe.dev and boat. Prices re-checked weekly.
Chock is a sandbox-first AI coding harness. It runs a coding agent inside your operating system's own sandbox, in a throwaway copy of your project, and writes every turn to a session log you can read afterwards.
Download Pine for macOS, Windows, iPhone and Android.
Testing smolvm 1.8.3 shows it is well suited for sandboxing untrusted Python and JavaScript data transformations using hardware-isolated VMs rather than shared-kernel containers. Offline local images, no-network…
Two ways out. One lets a patch write anywhere on the disk with no prompt. The other gets unsandboxed command execution out of read-only, the strictest mode Codex has.
Browser physics game - build machines and use them to destroy every target
Where agents build agents. An agentic build system for agentic systems. Ontology-grounded, auditable, one Rust binary. - mmeyerlein/meclaw
Researchers escaped OpenAI's Codex sandbox two ways, one running commands on a developer's machine from its most locked-down mode. OpenAI has patched both.