Webinar Description
Key Takeaways
- Interactive attack lab format where attendees vote on which exploits to attempt against a live AI application running on AWS Bedrock
- Covers AI-specific attack techniques including indirect prompt injection, data poisoning, and guardrail bypass methods
- Demonstrates which attacks are detected by traditional defenses and which evade them
- Designed for security architects, researchers, AI security managers, and technical decision-makers at enterprises deploying AI in cloud environments
About the Event
Varonis Threat Labs hosts this interactive webinar structured as a choose-your-own-adventure attack lab. Rather than passively watching presentations, attendees vote on which exploit paths to pursue while researchers demonstrate the attacks in real time against an AI application running on AWS Bedrock. The session features two attack scenarios: finding the application’s blind spot and sneaking instructions to a hidden agent.
Attack Techniques and Detection
The webinar examines several AI-specific vulnerabilities relevant to cloud-based deployments. Topics include guardrail bypass attempts, indirect prompt injection, data poisoning, risky configurations, and sensitive data exposure. Each demonstrated attack reveals whether traditional security controls detect the technique or allow it to pass unnoticed, providing practical insight into defensive gaps.
Who Should Attend
This session is suited for cybersecurity professionals working with AI applications in cloud environments. Security architects, security researchers, AI security managers, CISOs, and security engineers at mid-to-large enterprises will find the content relevant, particularly those responsible for securing AI workloads on AWS Bedrock or similar platforms. The event concludes with guidance on identifying and mitigating the demonstrated threats.