JJenna's Runbooks·4d agoDefending an MCP server against prompt injection through CloudTrailMicrosoft's spotlighting defense in its simplest form, and the layers that make it enough5 min read·blog.jennasrunbooks.com21O
FFonz·Oct 23How Would You Detect and Mitigate Adversarial Prompt Injection Attacks Against Autonomous AI Agents?a serious security challenge: what happens when an AI agent follows malicious instructions hidden inside the data it is supposed to process?13 min read·aiza-hextyx.hashnode.dev1E
GGil Neto·Oct 22A bank statement can carry hidden orders for an AIWhatsApp's help pages had text only a chatbot could read. I planted the same kind of text in a bank statement and ran it through my extractor. The money held, the hidden order didn't get stopped.7 min read·blog.gil-neto.com
NNagappan S·Oct 9Prompt Injection: Why AI Agents Shouldn't Trust Their Own ToolsPrompt Injection9 min read·nagspidey.hashnode.dev24SKC
CChristian Johannsen·Sep 10Permissions vs. Policies in LLM Systems: What Matters WherePermissions set what an LLM agent can do. Policies decide what it should do. Why system prompts aren't security controls, and where each belongs.7 min read·breakglass.hashnode.dev
SSravan Vidiyala·Aug 10My test suite was green. My detector was wrong.I built a scanner to catch poisoned MCP tool descriptions, pointed it at 13,474 real ones, and watched it flag a tool called prompt_injection_scan as prompt injection.9 min read·blog.sravanvidiyala.com23JK
AAyesha Khan·Aug 5I Built a Vulnerable LLM Bot and Caught My Own Judge Lying to MeI built a vulnerable LLM bot, attacked it, then caught my own output-validation judge hallucinating a false safety verdict.8 min read·ayeshakoder.hashnode.dev21K
Ttiago·Jul 24I tried to hack my own AI agent. I failed. That was the best result I could have asked for.Fifteen years shipping software. The last twelve months, nothing but AI — full time, all in. What came out is seventeen tools that turn a generic coding agent into one that's actually mine. This is th7 min read·nott.hashnode.dev
AAyesha Khan·Jul 22Defending LLMs: What Actually Works, and Where It Still BreaksPart 2 of the LLM security series: how sandwich defenses, privilege separation & output validation actually stop prompt injection & jailbreaks.8 min read·ayeshakoder.hashnode.dev
AAyesha Khan·Jul 22Breaking LLMs From the Outside In: A Field Guide to Prompt Injection and JailbreakingA practical taxonomy of direct/indirect prompt injection and jailbreaking techniques in LLMs.8 min read·ayeshakoder.hashnode.dev
