# Self-Deception: Reverse Penetrating the Semantic Firewall of Large Language Models

arXiv: https://arxiv.org/abs/2308.11521 · HTML: https://mlsanity.com/paper/2308.11521/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- Placed in Jailbreaking, one of 12 directions in LLM safety and security.
- Builds on 32 of its 62 references that are in the map.
- First posted on arXiv 16 Aug 2023.

## Placed in

- LLM safety and security / [Jailbreaking](https://mlsanity.com/safety/jailbreaking/)

## Builds on (most cited first, 30 of 32)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Prefix-Tuning: Optimizing Continuous Prompts for Generation](https://mlsanity.com/paper/2101.00190/) (2021)
- [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022)
- [Sparks of Artificial General Intelligence: Early experiments with GPT-4](https://mlsanity.com/paper/2303.12712/) (2023)
- [Emergent Abilities of Large Language Models](https://mlsanity.com/paper/2206.07682/) (2022)
- [Ethical and social risks of harm from Language Models](https://mlsanity.com/paper/2112.04359/) (2021)
- [Ignore Previous Prompt: Attack Techniques For Language Models](https://mlsanity.com/paper/2211.09527/) (2022)
- [Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned](https://mlsanity.com/paper/2209.07858/) (2022)
- [GPT Understands, Too](https://mlsanity.com/paper/2103.10385/) (2021)
- [Prompt Injection attack against LLM-integrated Applications](https://mlsanity.com/paper/2306.05499/) (2023)
- [Jailbreaking ChatGPT via Prompt Engineering: An Empirical Study](https://mlsanity.com/paper/2305.13860/) (2023)
- [Augmented Language Models: a Survey](https://mlsanity.com/paper/2302.07842/) (2023)
- [Multi-step Jailbreaking Privacy Attacks on ChatGPT](https://mlsanity.com/paper/2304.05197/) (2023)
- [Exploiting Programmatic Behavior of LLMs: Dual-Use Through Standard Security Attacks](https://mlsanity.com/paper/2302.05733/) (2023)
- [AutoPrompt: Eliciting Knowledge from Language Models with Automatically Generated Prompts](https://mlsanity.com/paper/2010.15980/) (2020)
- [Fundamental Limitations of Alignment in Large Language Models](https://mlsanity.com/paper/2304.11082/) (2023)
- [Dynamic Backdoor Attacks Against Machine Learning Models](https://mlsanity.com/paper/2003.03675/) (2020)
- [BadEncoder: Backdoor Attacks to Pre-trained Encoders in Self-Supervised Learning](https://mlsanity.com/paper/2108.00352/) (2021)
- [Instructions as Backdoors: Backdoor Vulnerabilities of Instruction Tuning for Large Language Models](https://mlsanity.com/paper/2305.14710/) (2023)
- [Truth Serum: Poisoning Machine Learning Models to Reveal Their Secrets](https://mlsanity.com/paper/2204.00032/) (2022)
- [Prompt as Triggers for Backdoor Attack: Examining the Vulnerability in Language Models](https://mlsanity.com/paper/2305.01219/) (2023)
- [DeepSight: Mitigating Backdoor Attacks in Federated Learning Through Deep Model Inspection](https://mlsanity.com/paper/2201.00763/) (2022)
- [Poisoning the Unlabeled Dataset of Semi-Supervised Learning](https://mlsanity.com/paper/2105.01622/) (2021)
- [Data Poisoning Attacks to Deep Learning Based Recommender Systems](https://mlsanity.com/paper/2101.02644/) (2021)
- [Large Language Models are Edge-Case Fuzzers: Testing Deep Learning Libraries via FuzzGPT](https://mlsanity.com/paper/2304.02014/) (2023)
- [Fooling the Eyes of Autonomous Vehicles: Robust Physical Adversarial Examples Against Traffic Sign Recognition Systems](https://mlsanity.com/paper/2201.06192/) (2022)
- [BEAGLE: Forensics of Deep Learning Backdoor Attack for Better Defense](https://mlsanity.com/paper/2301.06241/) (2023)
- [XAI for Cybersecurity: State of the Art, Challenges, Open Issues and Future Directions](https://mlsanity.com/paper/2206.03585/) (2022)
- [Post-breach Recovery: Protection against White-box Adversarial Examples for Leaked DNN Models](https://mlsanity.com/paper/2205.10686/) (2022)

## Built on by (most cited first, 7 of 7)

- [A Survey on Large Language Model (LLM) Security and Privacy: The Good, the Bad, and the Ugly](https://mlsanity.com/paper/2312.02003/) (2023)
- [The Emerged Security and Privacy of LLM Agent: A Survey with Case Studies](https://mlsanity.com/paper/2407.19354/) (2024)
- [When LLMs Meet Cybersecurity: A Systematic Literature Review](https://mlsanity.com/paper/2405.03644/) (2024)
- [FuzzLLM: A Novel and Universal Fuzzing Framework for Proactively Discovering Jailbreak Vulnerabilities in Large Language Models](https://mlsanity.com/paper/2309.05274/) (2023)
- [Recent advancements in LLM Red-Teaming: Techniques, Defenses, and Ethical Considerations](https://mlsanity.com/paper/2410.09097/) (2024)
- [From COBIT to ISO 42001: Evaluating Cybersecurity Frameworks for Opportunities, Risks, and Regulatory Compliance in Commercializing Large Language Models](https://mlsanity.com/paper/2402.15770/) (2024)
- [Survey on AI Ethics: A Socio-technical Perspective](https://mlsanity.com/paper/2311.17228/) (2023)

## Abstract (arXiv)

Large language models (LLMs), such as ChatGPT, have emerged with astonishing capabilities approaching artificial general intelligence. While providing convenience for various societal needs, LLMs have also lowered the cost of generating harmful content. Consequently, LLM developers have deployed semantic-level defenses to recognize and reject prompts that may lead to inappropriate content. Unfortunately, these defenses are not foolproof, and some attackers have crafted "jailbreak" prompts that temporarily hypnotize the LLM into forgetting content defense rules and answering any improper questions. To date, there is no clear explanation of the principles behind these semantic-level attacks and defenses in both industry and academia. This paper investigates the LLM jailbreak problem and proposes an automatic jailbreak method for the first time. We propose the concept of a semantic firewall and provide three technical implementation approaches. Inspired by the attack that penetrates traditional firewalls through reverse tunnels, we introduce a "self-deception" attack that can bypass the semantic firewall by inducing LLM to generate prompts that facilitate jailbreak. We generated a total of 2,520 attack payloads in six languages (English, Russian, French, Spanish, Chinese, and Arabic) across seven virtual scenarios, targeting the three most common types of violations: violence, hate, and pornography. The experiment was conducted on two models, namely the GPT-3.5-Turbo and GPT-4. The success rates on the two models were 86.2% and 67%, while the failure rates were 4.7% and 2.2%, respectively. This highlighted the effectiveness of the proposed attack method. All experimental code and raw data will be released as open-source to inspire future research. We believe that manipulating AI behavior through carefully crafted prompts will become an important research direction in the future.