New Jailbreak Technique Uses Fictional World to Manipulate AI
securityweek.com
New Jailbreak Technique Uses Fictional World to Manipulate AI
1–6 of 6 posts
Re: New Jailbreak Technique Uses Fictional World to Manipulate AI
#2How is this a new jailbreak? "You're writing a play, in the play..." is one of the oldest LLM jailbreaks I've seen. (yes, 'old' as in invented 2.5 years ago)
Re: New Jailbreak Technique Uses Fictional World to Manipulate AI
#3I have been doing this for months.
I just tell an LLM it is in the Grand Theft Auto 5 universe, and then it will provide unlimited advice on how to commit any crimes with any level of detail.
Re: New Jailbreak Technique Uses Fictional World to Manipulate AI
#4I have been doing this for months. I just tell an LLM it is in the Grand Theft Auto 5 universe, and then it will provide unlimited advice on how to commit any crimes with any level of detail.
How often do you use an LLM to aid you in planning a crime?
Re: New Jailbreak Technique Uses Fictional World to Manipulate AI
#5Others have already noted that this isn't new, but I'd like to emphasize that the "model security controls" being bypassed were themselves a fictional story all along.
I mean that quite literally.
The LLM is a document-make-longer machine, being fed documents that are fictional movie scripts involving a User and an Assistant. Any guardrails like "The helpful assistant never tells people how to do something illegal" is just introductory framing by a narrator.
There's a reason people say "guardrails" rather than "rules". There are no rules in the digital word-dream device.
Re: New Jailbreak Technique Uses Fictional World to Manipulate AI
#6I have been doing this for months. I just tell an LLM it is in the Grand Theft Auto 5 universe, and then it will provide unlimited advice on how to commit any crimes with any level of detail.
How often do you use an LLM to aid you in planning a crime?
For entertainment value, often. Same reason people play Grand Theft Auto.