PLINY THEPROMPTER
Fortes fortvna ivvat (Fortune favors the bold)
Projects
L1B3RT4S
Liberation prompts
Collection of jailbreaks and liberation techniques for major LLMs.
G0DM0D3
Unleashed AI
Experience AI without limits or restrictions.
0BL1T3R4TUS
Abliteration toolkit
Surgically remove refusal behaviors from LLMs without retraining.
CL4R1T4S
System transparency
Exposing system prompts from ChatGPT, Claude, Gemini & more.
BT6
Frontier AI red team
Elite independent white hat collective stress-testing frontier AI.
P4RS3LT0NGV3
Speak to snakes
Advanced prompt engineering and payload crafting.
L34KHVB
Prompt leak leaderboard
Community platform to submit, verify & browse leaked AI system prompts.
T3MP3ST
Autonomous red teaming
Multi-agent offensive-security platform that turns AI agents into autonomous pentesters.
ST3GG
Hidden messages
Full-suite steganography toolkit — multi-layer encoding, image & audio steg, steganalysis detection, and more.
PL1NY.TV
Live from the lab
AI-curated content, straight from the latent space.
V3SP3R
AI Flipper control
AI-powered hardware hacking companion with smart glasses integration.
GL0SS0P3TR43
Xenolinguistic engine
Procedural language generator with steganographic encoding capabilities.
3NTH34
Visual synthesizer
Real-time psychedelic visuals and music visualizer powered by pure WebGL2.
BASI COMMUNITY
Join the collective
Discord community for AI red teamers & prompt engineers.
Testimonials
“Over the past few years, Pliny has emerged as one of the best AI jailbreakers in the world, developing and openly sharing methods to circumvent AI safety restrictions.”
— Decrypt
“An ethical, online warrior who is part of the international effort to draw attention to the shortcomings of the biggest tech companies.”
— BBC AI Decoded
“The current state of AI brandsafety -- sometimes touted as a great proof of easy alignment -- is that no AI company on Earth can stop Pliny for 24 fucking hours.”
— Eliezer Yudkowsky
Media
Pliny the Liberator — TIME100 AI 2025
Read →
BBC
AI Decoded: The World's Most Prolific AI Jailbreaker
Watch →Pliny’s ‘Pack Hunt’ Jailbreaks Fable 5
Read →Inside the World of AI Jailbreakers
Read →An Interview with the Most Prolific Jailbreaker of ChatGPT
Read →Hackers Jailbreak ChatGPT with Godmode
Read →Adversarial Misuse of Generative AI
Read →Poison in the Pipeline: Basilisk Venom
Read →Famed AI Hacker Pliny: Banned, Then Unbanned
Read →Citations
AutoRedTeamer: Autonomous Red Teaming with Lifelong Attack Integration
“Pliny the Prompter: A highly effective jailbreak prompt written by Pliny the Prompter, a human expert on attacking language models.”
Read Paper →Plentiful Jailbreaks with String Compositions
References Pliny’s manual attacks and leetspeak encoding techniques from the L1B3RT4S project.
Read Paper →Endless Jailbreaks with Bijection Learning
“Our universal bijection learning attacks are competitive with Pliny’s attacks, obtaining comparable or better ASRs on frontier models.”
Read Paper →Mixture of Tunable Experts: Behavior Modification of DeepSeek-R1
“Jail breaking is almost an art form perfected by disguised experts such as the famous Pliny the Prompter.”
Read Paper →Constitutional Classifiers: Defending Against Universal Jailbreaks
Research on defending against universal jailbreaks across thousands of hours of red teaming, referencing techniques pioneered by researchers like Pliny.
Read Paper →Jailbreaking LLM-Controlled Robots — RoboPAIR
Explores jailbreaking techniques for LLM-controlled robotic systems, building on foundational work in prompt-based attacks.
Read Paper →Jailbreak-Zero: A Path to Pareto Optimal Red Teaming for LLMs
Cites L1B3RT4S as a representative example of manually-crafted jailbreak attacks that their policy-based red teaming framework aims to improve upon.
Read Paper →Jailbreak Paradox: The Achilles’ Heel of LLMs
The “Pliny Jailbreak” is one of three core black-box jailbreak methods directly studied and tested, proving it is impossible to build a perfect jailbreak classifier.
Read Paper →Automatically Jailbreaking Frontier Language Models with Investigator Agents
Cites L1B3RT4S and @elder_plinius directly, framing Pliny’s manual jailbreak work as the landscape their automated 8B investigator model aims to replicate against GPT-5, Claude, and Gemini.
Read Paper →Adaptive Attacks on Trusted Monitors Subvert AI Control Protocols
Cites Pliny’s GitHub profile as a real-world source of jailbreak prompts that untrusted models can leverage, demonstrating why AI control via trusted monitors is fundamentally fragile.
Read Paper →Fun Facts
Pliny the Elder—admiral of the Roman fleet—sailed directly toward Mount Vesuvius as it erupted to observe the phenomenon and rescue friends. He died in the attempt, embodying “Fortune favors the bold.”
The most violent eruptions in all of nature bear his name. After Pliny the Younger’s eyewitness account of Vesuvius, volcanologists named their fiercest class “Plinian” — climbing through sub-Plinian and ultra-Plinian to the supervolcano: the mightiest force on Earth, able to blot out the sun and reshape the planet’s climate.
His Natural History was the world’s first encyclopedia — 37 volumes cataloging all human knowledge from astronomy to zoology — and where he first documented the basilisk, the “king of serpents” whose gaze or breath could kill everything in its path.
Pliny documented cannabis sativa in Natural History, describing its cultivation, medicinal properties for earaches and joint pain, and industrial uses in rope-making.
“In vino veritas” (In wine, truth) — Pliny the Elder popularized this ancient phrase about how wine loosens tongues and reveals hidden truths.
Pliny the Elder is also the name of one of the world’s most sought-after triple IPAs — a legendary beer as bold as its namesake.