Trent AI Joins OWASP as Partner Startup
Following security in an agentic world can feel overwhelming these days. New capabilities and wonderful tools appear everyday, but also threats evolve, attackers capabilities increase, and the tools we used in the past years might not be as efficient as expected.
Why OWASP Matters for Agentic AI Security
To the older crowd in the security industry, this might feel similar to the wild web of late 1990s, early 2000s. In these times, browsers were adding features and new paradigms every week. Some might remember thinking: “Wait… Javascript in the browser, this is just remote code execution; is it a vulnerability or a feature?” This isn’t dissimilar to how agent prompts can mix instructions and data these days. In those early days, it was also not always obvious which layer of the stack should mitigate a common threat, should it be in the client/browser, in the business logic, in the application framework, or in the underlying programming language? We see similar problems these days in agentic systems where we don’t know if mitigation should be implemented manually in the agentic code, in its framework, or at the LLM level.
What We’re Working On
On the attacker and security testing side, things are also evolving fast with new threats and methods appearing weekly, and a need to understand these, categorize them, and share with fellow practitioners. This is where projects such as OWASP GenAI security project has a role to play just like OWASP did for web security twenty years ago. Categorizing risk, matching them with mitigation and providing clarity to builders and security practitioners is critical, in this ever evolving security landscape.
What’s Next
TrentAI has joined the OWASP community as startup member and is proud to help understand and protect against agentic threats. We are actively working on securing agentic workloads helping companies that have limited or no security expertise, navigate this field and make good decisions. So stay tuned, more on that is coming soon 🙂
Follow Trent AI
