Security Plugin for AI Agents Launched
Adversa AI has launched SecureClaw, which it describes as the first open-source security plugin aligned with OWASP standards for OpenClaw AI agents. The tool is designed to provide a security skill for autonomous AI systems. The announcement was made via a corrected press release that included updated hyperlinks.
- OpenClaw is an open-source AI agent framework created by Austrian developer Peter Steinberger that enables large language models to execute tasks on a user's behalf, such as sending emails, scraping websites, and managing files. The framework runs locally, connecting with messaging apps like WhatsApp, Telegram, and Slack to receive commands in natural language. - Adversa AI, the company behind SecureClaw, is an Israeli startup specializing in AI security. The company focuses on "AI Red Teaming" and securing autonomous AI agents, and has been recognized by Gartner for its work in the field. - SecureClaw's alignment with the Open Web Application Security Project (OWASP) Top 10 for Large Language Models addresses critical vulnerabilities in AI systems. These vulnerabilities include prompt injection, insecure output handling, and training data poisoning. - The SecureClaw plugin provides 55 automated audit and hardening checks for OpenClaw agents. It is designed to offer protection against known vulnerabilities and threats, and its security mappings are aligned with standards from MITRE ATLAS and the Cloud Security Alliance. - The launch of SecureClaw comes as interest in agentic AI grows, with major tech companies like Google and Microsoft developing their own AI agent capabilities. OpenClaw itself has seen rapid adoption in the developer community, becoming one of the most-starred repositories on GitHub since its launch in late 2025. - A recent report from Adversa AI highlighted that over 35% of real-world AI security incidents were caused by prompt-based exploits, underscoring the need for security solutions like SecureClaw.
