Israeli Startup AIR Security Develops Firewall for AI Agents
Israeli cybersecurity startup AIR Security has developed a specialized protection system designed to secure AI agents. The company announced it has raised $50 million for commercial development.
AIR Security's developers identify the content AI agents interact with as the primary threat. When AI agents process emails, browse websites, or utilize plugins and servers, they can encounter malicious instructions that manipulate their decision-making, potentially leading to data leaks or unauthorized access.
The AIR Security platform monitors and assesses various tools and extensions before and after they are integrated into a company's workflows. If a threat is detected, the system can block the dangerous module and revoke its use across all connected processes. The startup is also building a catalog of pre-vetted and certified add-ons.
Their research has identified over 17,000 public extensions linked to suspicious sources and instances where malicious applications impersonated well-known brands like OpenAI and Anthropic. The company emphasizes real-time, preventive protection and input filtering.
"Every company has a firewall to protect its network. Now everyone needs a firewall to protect their AI agents. AI agents need this type of protection that shields them from what enters their context," stated AIR Security co-founder and CEO Yair Saban. The company, founded in February 2026, secured its $50 million funding round led by Sequoia Capital.
