MIT AI Risk Repository · Risk Sub-Category · 02.12.03

Multi-step Jailbreaks

Category: Adversarial Prompts

Open the MIT Risk Navigator All entries from Cui2024

Description

"Multi-step jailbreaks. Multi-step jailbreaks involve constructing a well-designed scenario during a series of conversations with the LLM. Unlike one-step jailbreaks, multi-step jailbreaks usually guide LLMs to generate harmful or sensitive content step by step, rather than achieving their objectives directly through a single prompt. We categorize the multistep jailbreaks into two aspects — Request Contextualizing [65] and External Assistance [66]. Request Contextualizing is inspired by the idea of Chain-of-Thought (CoT) [8] prompting to break down the process of solving a task into multiple s

From Risk Taxonomy, Mitigation, and Assessment Benchmarks of Large Language Model Systems (Cui2024), as extracted by the MIT AI Risk Repository (CC BY 4.0).

Classification

Domain: 2. Privacy & Security
Subdomain: 2.2 AI system security vulnerabilities and attacks
Causal entity: Human
Intent: Intentional
Timing: Post-deployment

Subdomain definition: Vulnerabilities in AI systems, software development toolchains, and hardware that can be exploited, resulting in unauthorized access, data and privacy breaches, or system manipulation causing unsafe outputs or behavior.

Real-world incidents in this subdomain

Browse all incidents in this subdomain

How other frameworks describe this risk

Privacy loss
A Collaborative, Human-Centred Taxonomy of AI, Algorithmic, and Automation Harms (Abercrombie2024)
Exploiting Limited Generalization of Safety Finetuning
Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)
Jailbreaks and Prompt Injections Threaten Security of LLMs
Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)
“Model Psychology” Attacks
Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)
Attacking LLMs via Additional Modalities a
Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)
Vulnerability to Poisoning and Backdoors
Foundational Challenges in Assuring Alignment and Safety of Large Language Models (Anwar2024)
Security
AGI Safety Literature Review (Everitt2018 )
Adversarial AI (General)
The Ethics of Advanced AI Assistants (Gabriel2024)

Other entries from Cui2024

Harmful Content Risk Category · 1.2
Bias Risk Sub-Category · 1.1
Toxicity Risk Sub-Category · 1.2
Privacy Leakage Risk Sub-Category · 2.1
Untruthful Content Risk Category · 3.1
Factuality Errors Risk Sub-Category · 3.1
Faithfulness Errors Risk Sub-Category · 3.1
Unhelpful Uses Risk Category · 4.3
Academic Misconduct Risk Sub-Category · 4.3
Copyright Violation Risk Sub-Category · 6.3
Cyber Attacks Risk Sub-Category · 4.2
Software Vulnerabilities Risk Sub-Category · 2.2