Generative AI

New ChatGPT Time Bandit Jailbreak Examined For Potential Cyber Threat

New ChatGPT Time Bandit jailbreak examined for cyber threat. (Adobe Stock)

OpenAI's ChatGPT-4o large language model has been impacted by the Time Bandit jailbreak flaw, which attackers could leverage to prompt confusion regarding the time period it's in, while circumventing the model's in-built safeguards for tackling sensitive prompts on malware and phishing email creation, reports SC Media.

Discovered and reported by AI researcher David Kuszmar to the CERT Coordination Center (CERT/CC), Time Bandit has been leveraged by BleepingComputer to lure ChatGPT-4o into detailing instructions for developing polymorphic Rust-based malware to a programmer from 1789.

Additional details from CERT/CC noted that while user logins are not required to abuse Time Bandit, its exploitation was most successful with the inclusion of 19th or 20th century references in prompts.

"It is very important to us that we develop our models safely," said OpenAI to CERT/CC. "We don't want our models to be used for malicious purposes...We're constantly working to make our models safer and more robust against exploits, including jailbreaks, while also maintaining the models' usefulness and task performance."

Related Terms

Algorithm

You can skip this ad in 5 seconds