BR
Public post
Bharath Reddy
Sep 2026 • linkedin
Post excerpt
Your AI agent doesn’t need to be “hacked” in the traditional sense. Sometimes, natural language is enough. A carefully framed request can influence an agent to call a tool. A poisoned RAG document can inject instructions without the user typing anything malicious. Two individually authorized tools can become dangerous when chained together. And repeated harmless-looking queries can slowly reveal sensitive information. That’s what makes the AI attack surface different from traditional application security. The risk…
Find the people and context behind this post
Sign up for Super Carl to explore relationship context, warm paths, and relevant opportunities.