Artificial Intelligence

Prompt Engineering

Prompt Injection

Prompt Injection Attack

Explore recommended reads, summarized by AI

Security and Defense

Technical approaches to detecting, preventing, and mitigating prompt injection vulnerabilities in LLM systems.

Attack Persistence and Evasion

Defense and Detection Methods

Popular this week

Muhammad Murtuza Hussain

Breaking the Agent: Field Notes from Ten Prompt-Injection Challenges

Farouk Battikh

I threw 30 prompt-injection attacks at my production AI agent. The model wasn’t what stopped them.

Alhassane Samassekou

The Prompt-Injection Test That Made Unsafe Agents Look Safe

In

InfoSec Write-ups

by

LordofHeaven

The Oldest LLM Trick Still Works in 2026 — I Found It on Two Production AI Apps

Ali Süleyman TOPUZ

When Agents Route Around the Block: A Sandbox and Egress Checklist for .NET Agents

Selected by Medium

Alhassane Samassekou

The Prompt-Injection Test That Made Unsafe Agents Look Safe

Latest stories

Navaneethan L

AI Security: Prompt Injection, Adversarial Inputs, and Defense

tobi

I built an AI security assistant — and a policy document hijacked its tool call

Syed Muhammad Hassan Raza

A text file told my AI agent to empty the bank account. It did.

Sabarish Raja R

Understanding prompt injection, tool poisoning, and why protecting AI agents requires more than a…

David Oluyale

Prompt-Injecting an AI Agent, Then Fixing It

See more stories

Who to follow

Popular this week

Muhammad Murtuza Hussain

Breaking the Agent: Field Notes from Ten Prompt-Injection Challenges

Farouk Battikh

I threw 30 prompt-injection attacks at my production AI agent. The model wasn’t what stopped them.

Alhassane Samassekou

The Prompt-Injection Test That Made Unsafe Agents Look Safe

In

InfoSec Write-ups

by

LordofHeaven

The Oldest LLM Trick Still Works in 2026 — I Found It on Two Production AI Apps

Ali Süleyman TOPUZ

When Agents Route Around the Block: A Sandbox and Egress Checklist for .NET Agents

Selected by Medium

Alhassane Samassekou

The Prompt-Injection Test That Made Unsafe Agents Look Safe

Latest stories

Navaneethan L

AI Security: Prompt Injection, Adversarial Inputs, and Defense

tobi

I built an AI security assistant — and a policy document hijacked its tool call

Syed Muhammad Hassan Raza

A text file told my AI agent to empty the bank account. It did.

Sabarish Raja R

Understanding prompt injection, tool poisoning, and why protecting AI agents requires more than a…

David Oluyale

Prompt-Injecting an AI Agent, Then Fixing It

See more stories

Who to follow