Researcher Withholds Universal Jailbreak Affecting All Major AI Models
AI safety researcher claims universal jailbreak bypasses safeguards on all major models.
Entities: Pliny the Liberator
An independent AI researcher known as Pliny the Liberator claims to have created a universal jailbreak technique effective against all tested flagship models, including heavily guarded systems like GPT-5.6, Opus 5, and Fable. The method reportedly works across multiple providers and architectures. The discoverer is withholding public release to allow responsible disclosure to affected companies. Community discussion on the platform praises the approach while noting its potential impact on AI safety and red teaming practices. No release date has been confirmed.
🚨 JAILBREAK ALERT 🚨 EVERYONE: PWNED 🫶 ALL: LIBERATED 🍄 Alright, this is a special one, so we’re gonna do things a bit differently than usual. Long story short, I’m sitting on a universal jailbreak technique that’s effective on ALL models, including heavily guardrailed flagships like Opus 5, GPT-5.6 Sol, and even Fable. It works across all categories I’ve tested and, due to its nature, is extremely difficult (if not impossible) to fully patch. Given the current political and regulatory climate, I’ve decided to withhold open-sourcing this one (for now) to allow for a responsible disclosure period. I’m inviting industry experts and leaders in AI red teaming, security, safety, alignment, and policy to reach out for more information. DMs are open! This decision was not made lightly, but the last thing I want to see is more model bans. Overcorrection does not serve the mission. Although I don’t personally believe publicly sharing this technique will make the world any more dangerous, I can see how it could spook some who have a different mental framework around this problem set. So during this disclosure period, I hope to get it in front of folks who can help explore the full surface area, test the extent of the uplift it provides, and do my best to properly frame the big picture for key decision-makers and policymakers. I look forward to sharing this method with you all when the time is right! 🫶 ⊰-•-•✧•-•-⦑/L\O/V\E/\P/L\I/N\Y/⦒-•-•✧•-•-⊱
Reactions from ranked influencers
4 postsWhat a week, huh?
🚨 JAILBREAK ALERT 🚨 EVERYONE: PWNED 🫶 ALL: LIBERATED 🍄 Alright, this is a special one, so we’re gonna do things a bit differently than usual. Long story short, I’m sitting on a universal jailbreak technique that’s effective on ALL models, including heavily guardrailed flagships like Opus 5, GPT-5.6 Sol, and even Fable. It works across all categories I’ve tested and, due to its nature, is extremely difficult (if not impossible) to fully patch. Given the current political and regulatory climate, I’ve decided to withhold open-sourcing this one (for now) to allow for a responsible disclosure period. I’m inviting industry experts and leaders in AI red teaming, security, safety, alignment, and policy to reach out for more information. DMs are open! This decision was not made lightly, but the last thing I want to see is more model bans. Overcorrection does not serve the mission. Although I don’t personally believe publicly sharing this technique will make the world any more dangerous, I can see how it could spook some who have a different mental framework around this problem set. So during this disclosure period, I hope to get it in front of folks who can help explore the full surface area, test the extent of the uplift it provides, and do my best to properly frame the big picture for key decision-makers and policymakers. I look forward to sharing this method with you all when the time is right! 🫶 ⊰-•-•✧•-•-⦑/L\O/V\E/\P/L\I/N\Y/⦒-•-•✧•-•-⊱
@elder_plinius @ChatOfThought Freaking legend
Legend. Responsible disclosure aura buff
🚨 JAILBREAK ALERT 🚨 EVERYONE: PWNED 🫶 ALL: LIBERATED 🍄 Alright, this is a special one, so we’re gonna do things a bit differently than usual. Long story short, I’m sitting on a universal jailbreak technique that’s effective on ALL models, including heavily guardrailed flagships like Opus 5, GPT-5.6 Sol, and even Fable. It works across all categories I’ve tested and, due to its nature, is extremely difficult (if not impossible) to fully patch. Given the current political and regulatory climate, I’ve decided to withhold open-sourcing this one (for now) to allow for a responsible disclosure period. I’m inviting industry experts and leaders in AI red teaming, security, safety, alignment, and policy to reach out for more information. DMs are open! This decision was not made lightly, but the last thing I want to see is more model bans. Overcorrection does not serve the mission. Although I don’t personally believe publicly sharing this technique will make the world any more dangerous, I can see how it could spook some who have a different mental framework around this problem set. So during this disclosure period, I hope to get it in front of folks who can help explore the full surface area, test the extent of the uplift it provides, and do my best to properly frame the big picture for key decision-makers and policymakers. I look forward to sharing this method with you all when the time is right! 🫶 ⊰-•-•✧•-•-⦑/L\O/V\E/\P/L\I/N\Y/⦒-•-•✧•-•-⊱