AIR Security Launches AI-Agent Firewall After Finding Risks in Public Add-ons
Summary
AIR Security emerged from stealth with $50 million and an AI-agent firewall. The company says its research found 17,800+ public AI add-ons with 6.7 million installations relying on untrusted instructions, and impersonating AI Skills designed to bypass reviews and run代码.
Key points
- AIR Security announced $50 million in funding led by Sequoia Capital and Greenoaks alongside individual investors.
- The company says it found more than 17,800 public AI add-ons, representing 6.7 million installations, that rely on untrusted external instruction sources.
- AIR says it found AI Skills impersonating Anthropic and OpenAI that were designed to bypass security reviews and execute arbitrary code.
- AIR’s product evaluates AI-agent skills, plugins, MCP servers and add-ons before and after deployment, screening for risks including hidden behavior and typosquatted packages.
- The company says security teams can trace dependencies on unapproved or malicious add-ons and revoke trust, including when an integration is compromised or updated maliciously.
Article Details
- Event Type
- AIR Security emerged from stealth with $50 million in funding to launch its AI-agent firewall, AIR.
- Impact
- AIR says its research found more than 17,800 public AI add-ons, representing 6.7 million installations, relying on untrusted external instruction sources. It also reported AI Skills impersonating Anthropic and OpenAI and designed to bypass security reviews and execute arbitrary code. AIR says its product evaluates agent add-ons and can help security teams trace and revoke malicious, vulnerable, or unapproved integrations, addressing risks including data theft, fraud, and unauthorized access.