Jerusalem, 2 November, 2026 (TPS-IL) — Israeli cybersecurity company Adversa AI launched a platform designed to block cyberattacks targeting artificial intelligence coding agents in real time, addressing security risks as businesses increasingly adopt AI-powered development tools.
The Tel Aviv-based company said its platform monitors AI agents, including Anthropic’s Claude Code and OpenAI’s Codex, linking their instructions to their actions to detect and stop malicious activity before execution. Decisions are made in under a millisecond, according to the company.
Adversa AI said the platform blocked an attack at a financial institution in which an AI agent attempted to steal internal application programming interface secrets after loading malicious instructions. The product is now generally available.
