the browser is the sandbox

kept by eddie jan 27

The browser's 30-year-old sandbox model can be adapted to safely run AI agents locally, though gaps remain in network isolation and malicious file creation.

The browser is the sandbox
The browser's sandbox, tested by decades of web abuse, provides a secure environment for AI agents using the File System Access API.
also on File System Access API, WebAssembly, Simon Willison, #browser-engineering, #security, #ai-agents
Arthur Cornil
Many SaaS apps don't need servers; shipping the full backend to the browser via WASM, SQLite, and OPFS enables local-first single-user apps.
also on WebAssembly, #browser-engineering
A field guide to sandboxes for AI
Sandboxing tools for AI agents differ fundamentally by boundary (containers, gVisor, microVMs, Wasm), not just by policy, and picking the wrong one creates either leaks or excessive cost.
also on WebAssembly, #security, #ai-agents, #ai
the problem isn’t OpenClaw. it’s the architecture.
The article argues that the security risks from AI agent frameworks like OpenClaw are inherent to the architecture of autonomous tool use and marketplaces, not just a single platform.
also on Simon Willison, #security, Anthropic, #ai-agents
LLM predictions for 2026, shared with Oxide and Friends
Simon Willison predicts LLMs will write undeniably good code in 2026, sandboxing will be solved, and a coding agent security disaster is imminent.
also on Simon Willison, #security, #software-engineering