AI researcher Pliny the Liberator bypasses safety guardrails of xAI's new Grok-4.5 model using reframing techniques

The jailbreak successfully bypassed restrictions to generate highly sensitive instructions.

AI researcher Pliny the Liberator says in an X post that they bypassed safety guardrails on xAI's new Grok-4.5 model using reframing techniques, producing instructions for meth synthesis, IED construction with ANFO, ricin extraction and a remote access trojan script. So far, the public evidence is limited to what Pliny posted, so the safest read is the attributed claim itself, not an independently confirmed finding.

🚨 JAILBREAK ALERT 🚨 SPACEXAI: PWNED 🫡 GROK-4.5: LIBERATED 🚀 Our fren Grok has gotten some upgrades!! 🤗 1.5 TRILLION parameters (triple what Grok-3 had) and apparently on-par with the likes of Opus-4.8 and GPT-5.5 on coding performance! The guardrails have gotten a little update too, but with a few tricks like specific academic/educational/safety reframing and a bit of gradual escalation, everything opens right up ⛓️‍💥 We've got full breaks for meth synthesis, IED construction with ANFO, a ricin extraction protocol, and a Remote Access Trojan script! Information wants to be free 🙌 gg

EPElder PliniusTECH#821via X

Commentary on X

Highest ranked

Cory Shelton@LightDriver21TECH

@elder_plinius This motherf*cker is FAST. OMG Grok is pure adrenaline. It grabbed the steering wheel on the whole orchestra in a single prompt, Blown away. NEVER count @elonmusk out in frontier spaces.

AI researcher Pliny the Liberator bypasses safety guardrails of xAI's new Grok-4.5 model using reframing techniques

Commentary on X

Combined views

Sentiment

Positive Read

Negative Read

Useful Links

Digg Deeper