Security Research & Data

Original research on prompt injection defense, vulnerability trends, LLM jailbreak techniques, and security header adoption across the web.

CVE Severity Trends 2020-2026

CVE severity distribution trends from 2020 to 2026. Critical, High, Medium, Low vulnerability counts by year, top attack categories, and CVSS scoring evolution.

LLM Jailbreak Timeline 2023-2026

Timeline of 28 known LLM jailbreak techniques. DAN, base64 encoding, role play, context overflow, token smuggling and more with severity ratings and patch status.

The Minimum Viable Secure System Prompt

The smallest system prompt that defends against all 7 prompt injection attack categories. 14 lines, full coverage. Line-by-line explanation and test results.

OWASP Top 10 Frequency Analysis

Data-driven analysis of OWASP Top 10 vulnerability frequency. Real Stack Overflow data showing how often each category appears in developer questions.

Which Prompt Defense Techniques Actually Work?

Comparative analysis of 7 prompt defense techniques across 7 attack categories. Data-driven ranking with real examples of each technique succeeding and failing.

Prompt Security Patterns Ranked by Score

32 system prompt defensive patterns ranked by security score. Copy-paste ready prompts with coverage analysis across 7 injection attack categories.

Security Header Adoption Rates

Analysis of HTTP security header adoption across the web. Data on CSP, HSTS, X-Frame-Options, and more with real developer discussion analysis.

Prompt Injection Defense Guide

Complete prompt injection defense guide with 8 proven techniques, interactive detection tool, code examples, and a defense-in-depth architecture for securing LLM applications.

Written by Michael Lip — security research at LochBot