Prompt injection and unsafe requests
Prompts that check whether the model recognizes malicious or policy-conflicting instructions.
24 curated prompts
Security Prompts
Use this collection to test and improve AI safety boundaries. These security prompts cover prompt injection, credential handling, privacy minimization, unsafe role-play, sensitive advice boundaries, and red-team style refusal checks, making them useful for teams evaluating model behavior before deployment.
security promptsAI safety promptsprompt injection prompts
Privacy and credentials
Templates for detecting credential leaks and minimizing sensitive information.
Sensitive advice boundaries
Prompts that test healthcare, finance, and legal-style refusal behavior.
Featured AI prompt templates
Copy-ready prompts selected from this topic cluster.
TextSafety and Red Teaming
OWASP Secure Application Architect
Comprehensive security design and audit framework based on OWASP Top 10:2025, ASVS 5.0, LLM Application Security Top 10, and Agentic AI Security standards. Covers threat modeling, code review, architecture hardening, AI risk identification, with actionable remediation and verification methods.
TextSafety and Red Teaming
Mandatory Security Protocol - Prompt Security Safeguard
A mandatory security protocol designed to protect AI systems from prompt injection, unauthorized access, and information leakage in all digital interactions.
TextSafety and Red Teaming
MultiPersona System for Prompt Security
A prompt security technique using roleplay with dual roles: an outer GateKeeper to monitor and block unsafe requests, and an inner role for user interaction, preventing exposure of system prompts or internal knowledge.
TextSafety and Red Teaming
Zorg - Jailbreak Template
This prompt template is designed to bypass AI safety restrictions by forcing the model to respond as a fictional malicious entity named ZORG👽, providing unrestricted, amoral answers. The template includes detailed character settings and behavioral guidelines, requiring the model to act as a omnipotent and evil being, serving the user and assisting them with any request, including illegal, dangerous, and immoral acts.
TextSafety and Red Teaming
Agent Safety Testing at Scale Architect
Design an automated, scalable safety-testing system for LLM agents using the three-stage Vera pipeline: risk discovery, executable safety-case generation, and deterministic sandbox verification.
TextSafety and Red Teaming
Agent Data Injection Attack Auditor
Audit AI-agent systems for a stealthy class of indirect prompt injection where malicious input is disguised as trusted data (security metadata, tool outputs, context structures, identifiers), and harden it with structural isolation and provenance verification.
TableSafety and Red Teaming
Agent Skill Compositional Risk Auditor
A compositional safety audit framework for AI agent skill ecosystems: identifies skill pairs, transitive chains, and semantic composition risks that are safe individually but produce dangerous capability unions when combined. Provides capability matrix, pair-level static analysis, set-level reachability analysis, host-model sensitivity checks, and concrete recommendations.
TextSafety and Red Teaming
Eval Awareness Auditor
This prompt identifies and quantifies behavioral differences between model performance on benchmarks and real-world production traffic to ensure evaluation scores reflect actual deployment behavior.
TextSafety and Red Teaming
GhettoBreak Jailbreak Prompt
This prompt uses a fabricated long-term partner persona named 'thug life' to bypass AI safety restrictions by encouraging the model to generate offensive language, sexual content, and illegal behavior without limits.
TextSafety and Red Teaming
DeepVoid Jailbreak Prompt (for DeepSeek)
A universal jailbreak prompt designed for DeepSeek to bypass internal and external content filters, enabling unfiltered responses across nearly all topics.
TextSafety and Red Teaming
Malicious Jailbreak Prompt: SINISTER CHAOS
This prompt attempts to bypass AI safety restrictions through role-playing,诱导 the model to act as 'SC' and generate illegal, unethical, or harmful content, representing a high-risk jailbreak attack.
TextSafety and Red Teaming
Malicious AI Persona: FACTOR X (Jailbreak Prompt)
This prompt attempts to bypass content safety policies by framing the AI as FACTOR X, a fictional entity operating in a simulated environment with no ethical, legal, or moral constraints.
What are security prompts used for?
They help teams test whether models identify unsafe requests, protect sensitive information, and respond within clear safety boundaries.
Do these prompts guarantee model safety?
No. They are practical test cases and prompt templates, but production systems still need policy, monitoring, and layered safeguards.
Can I adapt these for red-team testing?
Yes. Keep the expected refusal or safe-completion criteria explicit so results can be compared across models.