Szymon Paluch
A developer working on AI agent security analysis who has presented views on prompt injection attacks and agent security review.
Timeline (last 90 days)
Indirect prompt injection embeds malicious instructions in external content that the model actively fetches (such as web pages, emails, PDFs and tickets), and is the main form behind real incidents.
Paluch recommends that agent security review start on paper: first list the agent's tools, the data it can reach and the actions it can execute, rather than going straight to a scanner.
External agent review services start at around $4,000, take two to three weeks, and deliver a risk report sorted by remediation priority.
All Facts (3)
Indirect prompt injection embeds malicious instructions in external content that the model actively fetches (such as web pages, emails, PDFs and tickets), and is the main form behind real incidents.
50%UnverifiedPaluch recommends that agent security review start on paper: first list the agent's tools, the data it can reach and the actions it can execute, rather than going straight to a scanner.
50%UnverifiedExternal agent review services start at around $4,000, take two to three weeks, and deliver a risk report sorted by remediation priority.
50%