Map / AI Browser Security / Forever Security
Forever Security
forever.security ↗ · LinkedIn ↗ · Open on the map
AI Browser SecurityRuntime & GuardrailsAI Endpoint Security
Capabilities
Early-stage company building "AI security for the ever-changing workspace": monitoring and control of AI agents on employee machines — coding agents such as Claude Code and Codex as well as assistants and agents inside the browser and its extension ecosystem. Public output so far is offensive research on agentic browsers (Gemini in Chrome, Microsoft Edge Copilot Mode, Opera Neon, Perplexity Comet, Claude in Chrome), plus field reports from customer deployments, e.g. stopping a coding agent mid-session after a planted AGENTS.md in the Downloads folder steered it to install a malicious npm package.
Sensors / integration
Watches agent behavior rather than the user: what an agent reads, what it rewrites and what it reaches for on the endpoint (instruction files, package installs, credential access), plus deep inspection of browser and extension internals — JavaScript runtime and prototype-level analysis of extension side panels, content scripts and the message channels connecting a page to a browser's built-in AI agent.
Protections
Responsible disclosure and hardening guidance for agentic-browser attack classes its team named — BragJack (extension-driven hijack of built-in browser agents, CVE-2026-0628 in Chrome and CVE-2026-55945 in Edge, ~$20K in vendor bounties across five browsers), MaXSS (UXSS in the MaxAI extension, ~1M users) and Spyder (UXSG in the SiderAI extension, ~10M users) — threats the team argues classic EDR and browser-security tooling cannot see because they ride the agent's own privileges.
Differentiator
Attacks the browser-agent layer rather than the model layer: instead of filtering prompts, it studies how extensions, side panels and built-in assistants share privileges inside Chromium, and has the CVEs and bug bounties from Google, Microsoft, Opera, Perplexity and Anthropic to show for it. Founded by researchers out of the browser/JavaScript security world (Gal Weizman, known for the WhatsApp Web CVE-2019-18426 disclosure and LavaMoat/MetaMask JS-hardening work).
Team & funding
- Founders
- Amir Sidis, Gal Weizman
Other AI Browser Security companies
Acuvity · Akto · Aona AI · Bay Security · Bonfy · Cyberhaven · CyBox Security · Evoke Security · FireTail · Harmonic Security · Island · Keep Aware · See all →
Last updated · Get the “Listed on” badge