GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher | Paper Citation Record

Observation f61039f2-9233-45ab-8844-f48251c0a235 · inbound

Low-Resource Languages Jailbreak GPT-4 cites this paper.

Low-Resource Languages Jailbreak GPT-4 GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 55

Resolution: verified exact; arxiv_id, observed 2026-05-17T09:24:14.144341Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 1e0f96ee-9d05-49d1-922f-9f53fb1f0d00 · inbound

AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models cites this paper.

AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 18

Resolution: verified exact; arxiv_id, observed 2026-05-12T16:28:04.036110Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 322b7129-6590-4282-a446-b73f7eaf3844 · inbound

Learning to Ask: When LLM Agents Meet Unclear Instruction cites this paper.

Learning to Ask: When LLM Agents Meet Unclear Instruction GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 34

Resolution: verified exact; arxiv_id, observed 2026-05-23T21:13:28.110741Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 17bb32fe-3f83-446e-9564-6c6cd57c1830 · inbound

DROJ: A Prompt-Driven Attack against Large Language Models cites this paper.

DROJ: A Prompt-Driven Attack against Large Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 16

Resolution: unresolved; no resolver link, observed 2026-08-12T21:05:18.303373Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 2a7ab1d8-fdaa-4c0b-8829-a7f4a30c3e39 · inbound

Does Safety Training of LLMs Generalize to Semantically Related Natural Prompts? cites this paper.

Does Safety Training of LLMs Generalize to Semantically Related Natural Prompts? GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 43

Resolution: unresolved; no resolver link, observed 2026-08-11T22:41:27.842458Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 50a81d2a-a775-4586-a196-c270c647630c · inbound

LIAR: Leveraging Inference Time Alignment (Best-of-N) to Jailbreak LLMs in Seconds cites this paper.

LIAR: Leveraging Inference Time Alignment (Best-of-N) to Jailbreak LLMs in Seconds GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 77

Resolution: unresolved; no resolver link, observed 2026-08-11T20:53:38.141085Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b7d80fef-5017-4c91-aaa1-1129a49dcf14 · inbound

Look Before You Leap: Enhancing Attention and Vigilance Regarding Harmful Content with GuidelineLLM cites this paper.

Look Before You Leap: Enhancing Attention and Vigilance Regarding Harmful Content with GuidelineLLM GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 29

Resolution: unresolved; no resolver link, observed 2026-08-11T18:53:15.424756Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 137eadbd-56b4-4846-a67d-8e4022145685 · inbound

JailPO: A Novel Black-box Jailbreak Framework via Preference Optimization against Aligned LLMs cites this paper.

JailPO: A Novel Black-box Jailbreak Framework via Preference Optimization against Aligned LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 38

Resolution: unresolved; no resolver link, observed 2026-08-11T11:17:49.113067Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 51edbe05-782d-44f1-9cf5-0b0c48e7f1b8 · inbound

DiffusionAttacker: Diffusion-Driven Prompt Manipulation for LLM Jailbreak cites this paper.

DiffusionAttacker: Diffusion-Driven Prompt Manipulation for LLM Jailbreak GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-11T05:31:31.710393Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b821d2a3-1ac4-4adc-a409-77e24fda3655 · inbound

LLM360 K2: Building a 65B 360-Open-Source Large Language Model from Scratch cites this paper.

LLM360 K2: Building a 65B 360-Open-Source Large Language Model from Scratch GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 155

Resolution: unresolved; no resolver link, observed 2026-08-10T20:54:02.524772Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 095ec9a2-ea40-4352-a2c7-27ea35e465f3 · inbound

Self-Instruct Few-Shot Jailbreaking: Decompose the Attack into Pattern and Behavior Learning cites this paper.

Self-Instruct Few-Shot Jailbreaking: Decompose the Attack into Pattern and Behavior Learning GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 48

Resolution: unresolved; no resolver link, observed 2026-08-10T20:35:52.881332Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation d1729677-0533-44d4-922b-4874f4d38f52 · inbound

LLMs are Vulnerable to Malicious Prompts Disguised as Scientific Language cites this paper.

LLMs are Vulnerable to Malicious Prompts Disguised as Scientific Language GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 51

Resolution: unresolved; no resolver link, observed 2026-08-10T15:26:44.981699Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 24215a34-9598-4fcd-a458-a9cae85f5bea · inbound

xJailbreak: Representation Space Guided Reinforcement Learning for Interpretable LLM Jailbreaking cites this paper.

xJailbreak: Representation Space Guided Reinforcement Learning for Interpretable LLM Jailbreaking GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 7

Resolution: unresolved; no resolver link, observed 2026-08-10T11:12:44.716762Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation c561d7ab-4da3-4796-bef4-5b3ed7ffc3bd · inbound

Jailbreaking LLMs' Safeguard with Universal Magic Words for Text Embedding Models cites this paper.

Jailbreaking LLMs' Safeguard with Universal Magic Words for Text Embedding Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 44

Resolution: unresolved; no resolver link, observed 2026-08-10T00:12:04.968909Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ed493799-d566-405d-aa80-3f587e1ba4f4 · inbound

Large Language Model Adversarial Landscape Through the Lens of Attack Objectives cites this paper.

Large Language Model Adversarial Landscape Through the Lens of Attack Objectives GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 2024

Resolution: unresolved; no resolver link, observed 2026-08-09T10:29:50.133416Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 42c9a326-b592-4675-b531-0c2f61ac1e75 · inbound

Safety Reasoning with Guidelines cites this paper.

Safety Reasoning with Guidelines GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 70

Resolution: unresolved; no resolver link, observed 2026-08-08T23:50:36.202058Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation f6012ba0-abe9-46d9-86b0-fa47570d8577 · inbound

Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety cites this paper.

Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 72

Resolution: verified exact; arxiv_id, observed 2026-05-23T04:42:33.940336Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation d770caef-559f-473c-ad9e-b1bdc2ad2504 · inbound

Jailbreaking to Jailbreak cites this paper.

Jailbreaking to Jailbreak GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 51

Resolution: unresolved; no resolver link, observed 2026-08-08T17:02:47.493991Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 8b269371-95d4-47cc-b59a-08fcfb34f3a5 · inbound

Amplified Vulnerabilities: Structured Jailbreak Attacks on LLM-based Multi-Agent Debate cites this paper.

Amplified Vulnerabilities: Structured Jailbreak Attacks on LLM-based Multi-Agent Debate GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 8

Resolution: unresolved; no resolver link, observed 2026-08-16T11:07:06.918876Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7eb6311d-6602-471b-b071-54c2cd139a17 · inbound

CipherBank: Exploring the Boundary of LLM Reasoning Capabilities through Cryptography Challenges cites this paper.

CipherBank: Exploring the Boundary of LLM Reasoning Capabilities through Cryptography Challenges GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 52

Resolution: unresolved; no resolver link, observed 2026-08-16T06:05:12.205312Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation ee39192c-dd86-474c-aa43-09fe0a7ec133 · inbound

ICL CIPHERS: Quantifying "Learning" in In-Context Learning via Substitution Ciphers cites this paper.

ICL CIPHERS: Quantifying "Learning" in In-Context Learning via Substitution Ciphers GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 64

Resolution: unresolved; no resolver link, observed 2026-08-16T05:58:34.090637Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation b45cb1cb-4d4b-4557-8dae-6afd2da702b0 · inbound

NeuRel-Attack: Neuron Relearning for Safety Disalignment in Large Language Models cites this paper.

NeuRel-Attack: Neuron Relearning for Safety Disalignment in Large Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 26

Resolution: unresolved; no resolver link, observed 2026-08-16T05:33:11.865614Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 25844ae7-1cff-4366-a129-75d7593c59d1 · inbound

Three Minds, One Legend: Jailbreak Large Reasoning Model with Adaptive Stacked Ciphers cites this paper.

Three Minds, One Legend: Jailbreak Large Reasoning Model with Adaptive Stacked Ciphers GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 38

Resolution: unresolved; no resolver link, observed 2026-08-07T15:08:13.720278Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 00eaeda2-fadc-4cff-8b0c-0c3331b6c350 · inbound

Lifelong Safety Alignment for Language Models cites this paper.

Lifelong Safety Alignment for Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 59

Resolution: unresolved; no resolver link, observed 2026-08-07T14:00:09.433962Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5c8b3860-2fa6-4cd5-a27d-726d7a764bc8 · inbound

Benign-to-Toxic Jailbreaking: Inducing Harmful Responses from Harmless Prompts cites this paper.

Benign-to-Toxic Jailbreaking: Inducing Harmful Responses from Harmless Prompts GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 61

Resolution: unresolved; no resolver link, observed 2026-08-07T14:01:11.295701Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 25ce4ff6-5ffe-4b86-864f-aa98df872c3b · inbound

From Hallucinations to Jailbreaks: Rethinking the Vulnerability of Large Foundation Models cites this paper.

From Hallucinations to Jailbreaks: Rethinking the Vulnerability of Large Foundation Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 24

Resolution: unresolved; no resolver link, observed 2026-08-07T12:34:35.156845Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9be9595b-16d4-4dec-87c0-bc76a894d4a5 · inbound

Adversarial Preference Learning for Robust LLM Alignment cites this paper.

Adversarial Preference Learning for Robust LLM Alignment GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-07T12:35:23.203550Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 5cffd341-abfa-4495-a903-16bbe2c8b757 · inbound

SafeTy Reasoning Elicitation Alignment for Multi-Turn Dialogues cites this paper.

SafeTy Reasoning Elicitation Alignment for Multi-Turn Dialogues GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 24

Resolution: unresolved; no resolver link, observed 2026-08-07T12:04:47.378971Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 85f15e9a-2344-4c10-ad7f-c18cdf49a089 · inbound

Align is not Enough: Multimodal Universal Jailbreak Attack against Multimodal Large Language Models cites this paper.

Align is not Enough: Multimodal Universal Jailbreak Attack against Multimodal Large Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 9

Resolution: unresolved; no resolver link, observed 2026-08-07T11:51:30.970995Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 30ef75e3-d5d3-4d24-9092-8d1f164173d8 · inbound

The Scales of Justitia: A Comprehensive Survey on Safety Evaluation of LLMs cites this paper.

The Scales of Justitia: A Comprehensive Survey on Safety Evaluation of LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 134

Resolution: unresolved; no resolver link, observed 2026-08-07T10:17:27.004439Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 94e37423-7f20-4f0d-a15e-2c7c1724b829 · inbound

InfoFlood: Jailbreaking Large Language Models with Information Overload cites this paper.

InfoFlood: Jailbreaking Large Language Models with Information Overload GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-07T01:02:31.303381Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation df86a486-20dc-4ec7-a603-7be884eef3dc · inbound

Exploring the Secondary Risks of Large Language Models cites this paper.

Exploring the Secondary Risks of Large Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 52

Resolution: verified exact; arxiv_id, observed 2026-05-19T09:42:14.014272Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 1524d6e6-286a-43ba-93ef-3b2fef453951 · inbound

We Should Identify and Mitigate Third-Party Safety Risks in MCP-Powered Agent Systems cites this paper.

We Should Identify and Mitigate Third-Party Safety Risks in MCP-Powered Agent Systems GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 93

Resolution: unresolved; no resolver link, observed 2026-08-07T00:32:36.731379Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 820d8f86-6714-42f3-8fd9-c2763326728f · inbound

Cross-Modal Obfuscation for Jailbreak Attacks on Large Vision-Language Models cites this paper.

Cross-Modal Obfuscation for Jailbreak Attacks on Large Vision-Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 35

Resolution: unresolved; no resolver link, observed 2026-08-06T23:42:42.971360Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7d626332-ac7c-4139-a90b-8b60aa8ecab5 · inbound

SEALGuard: Safeguarding the Multilingual Conversations in Southeast Asian Languages for LLM Software Systems cites this paper.

SEALGuard: Safeguarding the Multilingual Conversations in Southeast Asian Languages for LLM Software Systems GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 56

Resolution: unresolved; no resolver link, observed 2026-08-06T18:28:46.286783Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a938959c-9abd-4204-8835-ff1968fc4e5d · inbound

Paper Summary Attack: Jailbreaking LLMs through LLM Safety Papers cites this paper.

Paper Summary Attack: Jailbreaking LLMs through LLM Safety Papers GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 39

Resolution: unresolved; no resolver link, observed 2026-08-06T16:28:53.451033Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 3b26ea39-a295-475d-8cbc-ed3f3f0b5a75 · inbound

Jailbreaking Large Language Diffusion Models: Revealing Hidden Safety Flaws in Diffusion-Based Text Generation cites this paper.

Jailbreaking Large Language Diffusion Models: Revealing Hidden Safety Flaws in Diffusion-Based Text Generation GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 49

Resolution: unresolved; no resolver link, observed 2026-08-15T18:00:26.732311Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation af484381-9d39-4ae6-8e77-2f660dd1f53c · inbound

PUZZLED: Jailbreaking LLMs through Word-Based Puzzles cites this paper.

PUZZLED: Jailbreaking LLMs through Word-Based Puzzles GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 22

Resolution: unresolved; no resolver link, observed 2026-08-06T05:43:39.612853Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 7a546f1b-868d-443b-9131-adc1338483e6 · inbound

Layer-Wise Perturbations via Sparse Autoencoders for Adversarial Text Generation cites this paper.

Layer-Wise Perturbations via Sparse Autoencoders for Adversarial Text Generation GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 72

Resolution: unresolved; no resolver link, observed 2026-08-05T20:31:39.676177Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation a0dae1fc-7da4-42d3-9976-ed7b99e495be · inbound

Mitigating Jailbreaks with Intent-Aware LLMs cites this paper.

Mitigating Jailbreaks with Intent-Aware LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 46

Resolution: unresolved; no resolver link, observed 2026-08-15T17:29:10.169290Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 81dec6fa-9497-4c47-b32e-c685f32f3f44 · inbound

On Surjectivity of Neural Networks: Can you elicit any behavior from your model? cites this paper.

On Surjectivity of Neural Networks: Can you elicit any behavior from your model? GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 110

Resolution: unresolved; no resolver link, observed 2026-08-05T16:00:52.279969Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 9f740cc8-976e-4e1b-bdac-fa4eb26f8472 · inbound

GUARD: Guideline Upholding Test through Adaptive Role-play and Jailbreak Diagnostics for LLMs cites this paper.

GUARD: Guideline Upholding Test through Adaptive Role-play and Jailbreak Diagnostics for LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 33

Resolution: metadata mismatch; arxiv_id, observed 2026-05-18T21:36:52.497769Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation f96a562b-0d45-4976-86b5-95dfd84493ac · inbound

The Resurgence of GCG Adversarial Attacks on Large Language Models cites this paper.

The Resurgence of GCG Adversarial Attacks on Large Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 36

Resolution: unresolved; no resolver link, observed 2026-08-05T13:42:44.381806Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 0ad7cd4e-8ace-4db5-9746-3cd4c5e513c3 · inbound

Embedding Poisoning: Bypassing Safety Alignment via Embedding Semantic Shift cites this paper.

Embedding Poisoning: Bypassing Safety Alignment via Embedding Semantic Shift GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 42

Resolution: unresolved; no resolver link, observed 2026-08-15T16:21:58.960568Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation e0e95381-fe18-4746-a919-30ebdb36348b · inbound

Harmful Prompt Laundering: Jailbreaking LLMs with Abductive Styles and Symbolic Encoding cites this paper.

Harmful Prompt Laundering: Jailbreaking LLMs with Abductive Styles and Symbolic Encoding GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 6

Resolution: unresolved; no resolver link, observed 2026-08-04T17:27:25.863629Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 14e021f4-b94e-4ddc-addd-d14cec6975ed · inbound

When Smiley Turns Hostile: Interpreting How Emojis Trigger LLMs' Toxicity cites this paper.

When Smiley Turns Hostile: Interpreting How Emojis Trigger LLMs' Toxicity GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 54

Resolution: unresolved; no resolver link, observed 2026-08-04T17:09:24.212002Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 43b20aaf-a445-4c4a-8f84-a697768cd714 · inbound

Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts cites this paper.

Red-Bandit: Test-Time Adaptation for LLM Red-Teaming via Bandit-Guided LoRA Experts GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 47

Resolution: verified exact; arxiv_id, observed 2026-05-21T20:54:21.564674Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation c59c361b-10ba-4e26-8941-72637653d8ee · inbound

SecureWebArena: A Holistic Security Evaluation Benchmark for LVLM-based Web Agents cites this paper.

SecureWebArena: A Holistic Security Evaluation Benchmark for LVLM-based Web Agents GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 62

Resolution: verified exact; arxiv_id, observed 2026-05-18T08:16:06.624171Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 9baeea80-9417-4dea-bfe7-8ccb1a57a5d6 · inbound

Guardian-as-an-Advisor: Advancing Next-Generation Guardian Models for Trustworthy LLMs cites this paper.

Guardian-as-an-Advisor: Advancing Next-Generation Guardian Models for Trustworthy LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 87

Resolution: verified exact; arxiv_id, observed 2026-05-11T06:46:34.844765Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 5140e315-47f6-4d25-9f5a-b37147cfd572 · inbound

TrajGuard: Streaming Hidden-state Trajectory Detection for Decoding-time Jailbreak Defense cites this paper.

TrajGuard: Streaming Hidden-state Trajectory Detection for Decoding-time Jailbreak Defense GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 44

Resolution: verified exact; arxiv_id, observed 2026-05-11T00:35:50.772336Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 14a35fdd-318c-4514-9c66-5ac8d4e0bb2d · inbound

Modeling LLM Unlearning as an Asymmetric Two-Task Learning Problem cites this paper.

Modeling LLM Unlearning as an Asymmetric Two-Task Learning Problem GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 3

Resolution: metadata mismatch; arxiv_id, observed 2026-05-10T11:40:18.978189Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 159dede1-ceaf-4d62-8314-a2c29bc928c4 · inbound

Ethics Testing: Proactive Identification of Generative AI System Harms cites this paper.

Ethics Testing: Proactive Identification of Generative AI System Harms GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 81

Resolution: verified exact; arxiv_id, observed 2026-05-11T14:56:07.935256Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation e937ce11-2df3-4700-ab24-536993c41fb5 · inbound

Attention Is Where You Attack cites this paper.

Attention Is Where You Attack GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 16

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T15:26:23.715009Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 0e6f93ff-4ae1-4c0d-83d4-582bcf692458 · inbound

Jailbroken Frontier Models Retain Their Capabilities cites this paper.

Jailbroken Frontier Models Retain Their Capabilities GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 3

Resolution: metadata mismatch; arxiv_id, observed 2026-05-11T15:26:09.757309Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 45fd79a9-dde3-49b0-b5d6-d309292020fb · inbound

Exposing LLM Safety Gaps Through Mathematical Encoding:New Attacks and Systematic Analysis cites this paper.

Exposing LLM Safety Gaps Through Mathematical Encoding:New Attacks and Systematic Analysis GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 12

Resolution: verified exact; arxiv_id, observed 2026-05-12T10:51:31.123072Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 08daedea-83f2-4830-9833-275470a56b54 · inbound

A Systematic Investigation of RL-Jailbreaking in LLMs cites this paper.

A Systematic Investigation of RL-Jailbreaking in LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 18

Resolution: verified exact; arxiv_id, observed 2026-05-11T01:40:52.578620Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 920ca7fe-d442-40cd-8bd1-6f12fb1373c8 · inbound

A Systematic Investigation of RL-Jailbreaking in LLMs cites this paper.

A Systematic Investigation of RL-Jailbreaking in LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 18

Resolution: verified exact; arxiv_id, observed 2026-07-01T13:35:46.496963Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation f3c9c04b-a4f8-45eb-aa48-d3ad2a6e10f4 · inbound

A Systematic Investigation of RL-Jailbreaking in LLMs cites this paper.

A Systematic Investigation of RL-Jailbreaking in LLMs GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 18

Resolution: unresolved; no resolver link, observed 2026-08-02T14:41:07.326795Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 78ba9926-cbda-44ab-b93d-dd356a0f8857 · inbound

From AI-Generated Content to Agentic Action: Security and Safety Threats in Generative AI cites this paper.

From AI-Generated Content to Agentic Action: Security and Safety Threats in Generative AI GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 151

Resolution: verified exact; arxiv_id, observed 2026-05-20T18:08:50.682304Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 00feaf7b-eb0f-48c4-afd2-5ae4f0ea2e8c · inbound

LASH: Adaptive Semantic Hybridization for Black-Box Jailbreaking of Large Language Models cites this paper.

LASH: Adaptive Semantic Hybridization for Black-Box Jailbreaking of Large Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 38

Resolution: verified exact; arxiv_id, observed 2026-05-21T04:49:35.515127Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 47ee0339-5dd1-488d-a45e-0832c8ee7588 · inbound

Adversarial Reframing: A Framework for Targeted Generation in Language Models cites this paper.

Adversarial Reframing: A Framework for Targeted Generation in Language Models GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 59

Resolution: metadata mismatch; arxiv_id, observed 2026-05-22T09:36:20.970501Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 41c38ef4-9f73-41ee-be10-84ed9c64bd34 · inbound

LCO: LLM-based Constraint Optimization for Safer Agentic LLMs in Real-world Tasks cites this paper.

LCO: LLM-based Constraint Optimization for Safer Agentic LLMs in Real-world Tasks GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 3

Resolution: malformed identifier; no resolver link, observed 2026-07-13T08:47:59.151393Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 141d8132-5853-478d-a121-92d13b184982 · inbound

SPARD: Defending Harmful Fine-Tuning Attack via Safety Projection with Relevance-Diversity Data Selection cites this paper.

SPARD: Defending Harmful Fine-Tuning Attack via Safety Projection with Relevance-Diversity Data Selection GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 19

Resolution: verified exact; arxiv_id, observed 2026-06-29T13:53:28.684601Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 55b51ddd-e3b9-44f3-9062-e9740a64b972 · inbound

Beyond English and Evasion: A Human-Annotated Multi-Domain Benchmark for High-Stakes LLM Safety Evaluation in Chinese cites this paper.

Beyond English and Evasion: A Human-Annotated Multi-Domain Benchmark for High-Stakes LLM Safety Evaluation in Chinese GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 14

Resolution: metadata mismatch; arxiv_id, observed 2026-06-29T08:03:14.660791Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation c5a5d93a-df35-44b3-85c6-8f4a8edbbbe0 · inbound

SlotGCG: Exploiting the Positional Vulnerability in LLMs for Jailbreak Attacks cites this paper.

SlotGCG: Exploiting the Positional Vulnerability in LLMs for Jailbreak Attacks GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 29

Resolution: metadata mismatch; arxiv_id, observed 2026-07-02T13:26:59.318146Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 769321d3-7363-4023-a416-b65a0668cf49 · inbound

Safety Paradox: How Enhanced Safety Awareness Leaves LLMs Vulnerable to Posterior Attack cites this paper.

Safety Paradox: How Enhanced Safety Awareness Leaves LLMs Vulnerable to Posterior Attack GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 5

Resolution: metadata mismatch; arxiv_id, observed 2026-07-02T12:36:56.906206Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation ba6e8dbd-d2da-40dc-9db7-14d0e308fd29 · inbound

FinRED: An Expert-Guided Benchmark Generation and Evaluation Framework for Financial LLM Red-Teaming cites this paper.

FinRED: An Expert-Guided Benchmark Generation and Evaluation Framework for Financial LLM Red-Teaming GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 45

Resolution: verified exact; arxiv_id, observed 2026-07-04T04:09:34.792750Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 57a91a59-846d-452b-86ec-c4b997c455f6 · inbound

An Empirical Evaluation of Prompt Injection Vulnerabilities in Large Language Models Across Multilingual and Obfuscated Attack Scenarios cites this paper.

An Empirical Evaluation of Prompt Injection Vulnerabilities in Large Language Models Across Multilingual and Obfuscated Attack Scenarios GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 24

Resolution: metadata mismatch; arxiv_id, observed 2026-06-30T06:54:20.370458Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 962a42cb-4562-4ad7-841a-b05e624492e8 · inbound

Mitigating Taint-Style Vulnerabilities in MCP Servers via Security-Aware Tool Descriptions cites this paper.

Mitigating Taint-Style Vulnerabilities in MCP Servers via Security-Aware Tool Descriptions GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 67

Resolution: verified exact; local_arxiv, observed 2026-07-09T10:26:11.075003Z

Source-reported events for the cited work

No event found in the named queried sources as of 2026-08-18T06:34:40.430872+00:00.

Observation 62146656-85ba-4c8c-84a4-d3c713cab06b · inbound

MJ: Multi-turn LLM Jailbreaking via Decomposed Credit Assignment cites this paper.

MJ: Multi-turn LLM Jailbreaking via Decomposed Credit Assignment GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 12

Resolution: unresolved; no resolver link, observed 2026-07-14T07:16:57.009797Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Observation 4fe603d7-2991-42d2-8292-8dd403e39b73 · inbound

Reason Before You Retrieve: Agentic Planning for Multi-modal RAG cites this paper.

Reason Before You Retrieve: Agentic Planning for Multi-modal RAG GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Cipher

Reference 110

Resolution: unresolved; no resolver link, observed 2026-08-02T10:20:56.070529Z

Source-reported events for the cited work

Unavailable: canonical work link unavailable.

Coverage vector

Reference resolution

Outbound references

Pith citing papers