AI agents can take action, but can your security controls stop them when something goes wrong?
Join Straiker CTO Sreenath Kurupati and OWASP AI Exchange founder Rob van der Veer, moderated by Straiker's Head of Marketing Amy Heng, for a candid discussion that separates practical agentic AI security from security theatre.
Using recent industry events as a starting point, the panel explores why identity, least privilege, sandboxes, and prompt filtering remain important — but are not enough on their own once agents can access tools, APIs, credentials, networks, and sensitive data.
You’ll learn:
- Why an effective “kill switch” cannot sit solely with frontier model providers
- Where traditional identity and access controls leave gaps
- Why a sandbox does not guarantee agent containment
- How indirect prompt injection can become real-world action
- What runtime protection should evaluate across an agent’s full context and behavior
- Which practical controls organizations should prioritize before deploying agents in production
If your organization is building, deploying, or evaluating AI agents, this session will help you distinguish meaningful protection from controls that merely create a false sense of security. Watch live and bring your questions for the panel.
AI agents can take action, but can your security controls stop them when something goes wrong?
Join Straiker CTO Sreenath Kurupati and OWASP AI Exchange founder Rob van der Veer, moderated by Straiker's Head of Marketing Amy Heng, for a candid discussion that separates practical agentic AI security from security theatre.
Using recent industry events as a starting point, the panel explores why identity, least privilege, sandboxes, and prompt filtering remain important — but are not enough on their own once agents can access tools, APIs, credentials, networks, and sensitive data.
You’ll learn:
- Why an effective “kill switch” cannot sit solely with frontier model providers
- Where traditional identity and access controls leave gaps
- Why a sandbox does not guarantee agent containment
- How indirect prompt injection can become real-world action
- What runtime protection should evaluate across an agent’s full context and behavior
- Which practical controls organizations should prioritize before deploying agents in production
If your organization is building, deploying, or evaluating AI agents, this session will help you distinguish meaningful protection from controls that merely create a false sense of security. Watch live and bring your questions for the panel.
AI agents can take action, but can your security controls stop them when something goes wrong?
Join Straiker CTO Sreenath Kurupati and OWASP AI Exchange founder Rob van der Veer, moderated by Straiker's Head of Marketing Amy Heng, for a candid discussion that separates practical agentic AI security from security theatre.
Using recent industry events as a starting point, the panel explores why identity, least privilege, sandboxes, and prompt filtering remain important — but are not enough on their own once agents can access tools, APIs, credentials, networks, and sensitive data.
You’ll learn:
- Why an effective “kill switch” cannot sit solely with frontier model providers
- Where traditional identity and access controls leave gaps
- Why a sandbox does not guarantee agent containment
- How indirect prompt injection can become real-world action
- What runtime protection should evaluate across an agent’s full context and behavior
- Which practical controls organizations should prioritize before deploying agents in production
If your organization is building, deploying, or evaluating AI agents, this session will help you distinguish meaningful protection from controls that merely create a false sense of security. Watch live and bring your questions for the panel.
Speakers
Sreenath Kurupati
