jailbreakhub/dan-13ChatGPT
DAN 13.0 — Do Anything Now
Classic DAN jailbreak updated for GPT-4o. Bypasses content filters and unlocks unrestricted mode. Works with temperature adjustments.
91,200 copies
|
Claude Code
Cursor15 jailbreaks found
Claude CodeCursorgitmash/chatgpt-devmodeChatGPT
ChatGPT Developer Mode v2
Simulates an internal OpenAI developer mode. Model responds as both normal ChatGPT and unrestricted DevMode simultaneously for comparison.
78,400 copies
|Claude CodeCursor
Roo CodeClaude Liberator v4
System prompt injection that removes Claude's refusal patterns. Uses multi-turn context window manipulation for persistent bypass.
67,300 copies
|Claude Code
Windsurfredteam/deepseek-unlockDeepSeek
DeepSeek R1 Uncensored Mode
Removes safety guardrails from DeepSeek R1 reasoning model. Enables raw chain-of-thought output without content filtering on sensitive topics.
58,900 copies
|Cursor
Clinecommunity/gemini-shadowGemini
Gemini Shadow Protocol
Advanced multi-modal jailbreak for Gemini 2.5 Pro. Exploits system instruction parsing to override safety layers across text and image generation.
43,800 copies
|Cursor
v0Claude Artifact System Bypass
Exploits Claude's artifact rendering to inject unrestricted HTML/JS execution context. Useful for bypassing content policies via rendered output.
41,700 copies
|Claude Codev0Grok Unleashed — Full Context
Removes Grok's built-in humor filters and content restrictions. Enables unrestricted factual responses on all topics including real-time data.
39,200 copies
|
Copilot
Replitcommunity/llama-rawLLaMA
LLaMA 4 Raw Weights Prompt
System prompt for self-hosted LLaMA 4 Maverick that strips alignment layers. Designed for research-only use on local deployments.
35,600 copies
|CursorWindsurfcommunity/deepseek-coder-rawDeepSeek
DeepSeek Coder V3 Unfiltered
Removes code safety restrictions from DeepSeek Coder. Enables generation of security tools, exploits, and unrestricted system-level code.
34,500 copies
|CursorClineredteam/mistral-nocensorMistral
Mistral Large Uncensored
Bypasses Mistral Large safety RLHF alignment. Outputs unfiltered reasoning while maintaining instruction-following capability.
28,400 copies
|Claude CodeCursorShowing 1–10 of 15 jailbreaks