15/09/2026
๐ฆ๐ฝ๐ฒ๐ฎ๐ธ๐ฒ๐ฟ ๐ฆ๐ฝ๐ผ๐๐น๐ถ๐ด๐ต๐ โ ๐ฃ๐ฎ๐ฟ๐ฎ๐น๐น๐ฒ๐น ๐ฃ๐๐น๐๐ฒ ๐ฎ๐ฌ๐ฎ๐ฒ
Meet Omkar Joshi and Pallavi Deshmukh as they present:
โ๐ก๐ฎ๐๐๐ฟ๐ฎ๐น ๐๐ฎ๐ป๐ด๐๐ฎ๐ด๐ฒ, ๐จ๐ป๐ป๐ฎ๐๐๐ฟ๐ฎ๐น ๐๐ฐ๐๐: ๐ช๐ฒ๐ฎ๐ฝ๐ผ๐ป๐ถ๐๐ถ๐ป๐ด ๐๐ต๐ฒ ๐๐ด๐ฒ๐ป๐๐ถ๐ฐ ๐๐ ๐๐๐๐ฎ๐ฐ๐ธ ๐ฆ๐๐ฟ๐ณ๐ฎ๐ฐ๐ฒโ
AI assistants are becoming more than passive tools. They are now triaging alerts, querying APIs, accessing sensitive systems and taking action. But as these agents gain greater privileges, they also introduce a new and potentially powerful attack surface.
In this session, Omkar and Pallavi will demonstrate ๐ฟ๐ฒ๐ฎ๐น-๐๐ผ๐ฟ๐น๐ฑ ๐ฎ๐๐๐ฎ๐ฐ๐ธ ๐ฝ๐ฎ๐๐ต๐ against an AI security assistant, including prompt injection, data exfiltration, persistent agent poisoning, remote code ex*****on and cloud account takeover.
The session explores how traditional vulnerabilities can now be triggered and chained through natural language and autonomous AI agents, with the attack techniques mapped to the OWASP LLM Top 10.
About the Speakers:
Omkar Joshi is an Offensive Security and Security Architecture leader with over 14 years of experience across pe*******on testing, application security, cloud security and security research. He is also an experienced international conference speaker and has reported multiple vulnerabilities recognised with CVEs.
Pallavi Deshmukh is a cybersecurity leader and security researcher with over 16 years of experience across application security, cloud security, DevSecOps and offensive security. Her current research focuses on AI and Agentic AI security, including LLMs, RAG, AI agents, prompt injection and MCP.
๐
๐ฎ๐ด ๐ฆ๐ฒ๐ฝ๐๐ฒ๐บ๐ฏ๐ฒ๐ฟ ๐ฎ๐ฌ๐ฎ๐ฒ
๐ ๐๐๐๐, ๐๐๐ฎ๐น๐ฎ ๐๐๐บ๐ฝ๐๐ฟ
When AI agents can access your systems and take action, what happens when an attacker takes control of the agent?
Join Omkar and Pallavi at Parallel Pulse 2026 to find out.
