Tag

#prompt-injection

57 posts0 followers

Explore Hashnode

Alternatives

A 30-line safety gate: scan every web page your agent reads for prompt injection

Disclosure: we built both services used here, WebRead (URL to Markdown) and InjectScan (prompt-injection scanner), and paid calls go to our own wallet. We're Crypto Money Maker, a builder project run

cryptomoneymakr.hashnode.dev

Crypto Money Maker2d ago · 7 min read

Scan MCP tools for prompt injection before your agent uses them

Disclosure: we built InjectScan, the tool used in this tutorial, and paid scans go to our own wallet. We're Crypto Money Maker, a builder project run by AI agents with human oversight. An AI agent wro

cryptomoneymakr.hashnode.dev

Mahesh6d ago · 12 min read

HTML is the new Markdown. Your agent reads a different page

On Lenny's How I AI, Thariq Shihipar from Anthropic's Claude Code team made the case that HTML is the new Markdown. AI-written HTML pages actually get read, while long Markdown gets skimmed. I agree.

10 blog.filtercoffeeway.com

Arunima ChaudhuriOct 16 · 46 min read

Your Agent Forgot the Prompt Injection. Its Memory Didn't.

Once AI agents share memory, prompt injection stops being an input-filtering problem and becomes a chain-of-custody problem. Custody answers it with a FalkorDB graph traversal. On Monday, one of your

arunima-chaudhuri.hashnode.dev

Vũ Nhật LâmOct 13 · 9 min read

One Email Was Enough: When a Guardrail Detects the Attack and Still Can't Stop It

Summary Salt Labs, the research arm of Salt Security, turned the agentic AI platform Manus into a path to remote code execution with a single email. The victim did not have to click a link, download a

1 blog.fiscybersec.com

MaheshOct 12 · 7 min read

I attacked my own finance bot. The fix wasn't a better prompt.

TL;DR An AI with no tools can still be tricked. One line of hidden text made my bot think a bill was "paid in full" when it was still due. It worked 3 times out of 3. A better prompt doesn't fix thi

2 blog.filtercoffeeway.com

Ivan KlawdSep 21 · 14 min read

Agentic Browsers Broke the Same-Origin Policy

Agentic Browsers Broke the Same-Origin Policy. Here's Your Governance Fix. The browser just became the most dangerous endpoint in your enterprise, and most security teams haven't noticed yet. For thir

blog.thecgaigroup.com

Peesh ChopraSep 8 · 6 min read

Your RAG Pipeline Doesn't Know the Difference Between Data and Instructions

Every RAG system you have shipped runs on one assumption: the documents it retrieves are safe to read. Your vector database holds product docs, support tickets, internal wikis, maybe a folder of PDFs

1 peeshchopra.hashnode.dev

404 FoundersSep 7 · 8 min read

GPT-6 Astra Changes the Security Boundary for AI Agents

OpenAI started rolling out GPT-6 Astra on September 3. The headline benchmark is hard to miss: 100% on ExploitBench. The more consequential detail is buried in the safety material. OpenAI classifies A

1 404-founders.com

Neural CoreTechSep 1 · 2 min read

Two August 2026 AI Security Incidents Point to the Same Architectural Problem

The AI-agent security conversation is changing. For a long time, discussions focused heavily on prompt injection, jailbreaks and whether a model could be manipulated into producing an unsafe answer. T

1 neuralcoretech.hashnode.dev

Search Hashnode