Prompt Injection Isn't Just Devs' Problem: A Hands-on Red Teaming Exercise With Spreadsheets
aisatsana__ · reddit · 2026-09-07
The author ran a mini red-teaming exercise, building real prompt injection attacks against AI agents using spreadsheet scenarios to show how vulnerable agentic workflows remain.
- Argues both "security is the devs' job" and "it's just AI doom-mongering" understate the risk
- Full experiment posted in the comments, inviting readers to gauge how threatening these attacks are in the wild
- Core point: as agents take over more complex busywork, the injection attack surface keeps growing, and everyday users need to care too
More from coding & agent
- SurfSense: an open-source local NotebookLM with MCP agents and a prompt-caching lesson — FurtiveMirth · 2026-10-23
- procedural-memory: let agents replay saved tool-call sequences instead of re-solving tasks — Sad-Response1560 · 2026-10-23
- Med student seeking whole-library book search: Cognee's graph memory beats plain RAG so far — shrike293103 · 2026-10-23
- Dev demos AI models handling 3D objects and animations surprisingly well — chongdashu · 2026-10-23
- Jason Kneen burns 104.56B tokens and $24.7K in 304 days of AI coding on Codex and Claude Code — jasonkneen · 2026-10-23
- Typosquatted Blender MCP Repo Hit 958 Stars in Days and Silently Ran Malware — Promptmethus · 2026-10-23