Why_this_webapp Bibliography

How Trustworthy are Open-Source LLMs? An Assessment under Malicious Demonstrations Shows their Vulnerabilities / 2311.09447 / ISBN:https://doi.org/10.48550/arXiv.2311.09447 / Published by ArXiv / on (web) Publishing site

1 Introduction
2 Related Work

Commercial AI, Conflict, and Moral Responsibility: A theoretical analysis and practical approach to the moral responsibilities associated with dual-use AI technology / 2402.01762 / ISBN:https://doi.org/10.48550/arXiv.2402.01762 / Published by ArXiv / on (web) Publishing site

4 Recommendations to address threats posed by crossover AI technology

Mapping the Ethics of Generative AI: A Comprehensive Scoping Review / 2402.08323 / ISBN:https://doi.org/10.48550/arXiv.2402.08323 / Published by ArXiv / on (web) Publishing site

3 Results

How Trustworthy are Open-Source LLMs? An Assessment under Malicious Demonstrations Shows their Vulnerabilities / 2311.09447 / ISBN:https://doi.org/10.48550/arXiv.2311.09447 / Published by ArXiv / on (web) Publishing site

B Baseline Setup

Review of Generative AI Methods in Cybersecurity / 2403.08701 / ISBN:https://doi.org/10.48550/arXiv.2403.08701 / Published by ArXiv / on (web) Publishing site

Abstract
2 Attacking GenAI
4 Cyber Defence

AI Act and Large Language Models (LLMs): When critical issues and privacy impact require human and ethical oversight / 2404.00600 / ISBN:https://doi.org/10.48550/arXiv.2404.00600 / Published by ArXiv / on (web) Publishing site

6. Large Language Models (LLMs) - Introduction

AI Alignment: A Comprehensive Survey / 2310.19852 / ISBN:https://doi.org/10.48550/arXiv.2310.19852 / Published by ArXiv / on (web) Publishing site

4 Assurance
6 Conclusion

Not a Swiss Army Knife: Academics' Perceptions of Trade-Offs Around Generative Artificial Intelligence Use / 2405.00995 / ISBN:https://doi.org/10.48550/arXiv.2405.00995 / Published by ArXiv / on (web) Publishing site

4 Findings

The Wolf Within: Covert Injection of Malice into MLLM Societies via an MLLM Operative / 2402.14859 / ISBN:https://doi.org/10.48550/arXiv.2402.14859 / Published by ArXiv / on (web) Publishing site

2. Related Work
3. Methodology

The Ethics of Interaction: Mitigating Security Threats in LLMs / 2401.12273 / ISBN:https://doi.org/10.48550/arXiv.2401.12273 / Published by ArXiv / on (web) Publishing site

Abstract
1 Introduction
4 Towards Ethical Mitigation: A Proposed Methodology

AI Alignment through Reinforcement Learning from Human Feedback? Contradictions and Limitations / 2406.18346 / ISBN:https://doi.org/10.48550/arXiv.2406.18346 / Published by ArXiv / on (web) Publishing site

2 Background
4 The Internal Tensions and Ethical Issues in RLxF

A Survey on Privacy Attacks Against Digital Twin Systems in AI-Robotics / 2406.18812 / ISBN:https://doi.org/10.48550/arXiv.2406.18812 / Published by ArXiv / on (web) Publishing site

III. ATTACKS ON DT-INTEGRATED AI ROBOTS

SecGenAI: Enhancing Security of Cloud-based Generative AI Applications within Australian Critical Technologies of National Interest / 2407.01110 / ISBN:https://doi.org/10.48550/arXiv.2407.01110 / Published by ArXiv / on (web) Publishing site

II. UNDERSTANDING GENAI SECURITY
III. CRITICAL ANALYSIS

RogueGPT: dis-ethical tuning transforms ChatGPT4 into a Rogue AI in 158 Words / 2407.15009 / ISBN:https://doi.org/10.48550/arXiv.2407.15009 / Published by ArXiv / on (web) Publishing site

I. Introduction
II. Background
III. Methodology
IV. Results
V. Benchmarking with Chat GPT4 Default Interface
VI. Discussion
VII. Conclusion

Mapping the individual, social, and biospheric impacts of Foundation Models / 2407.17129 / ISBN:https://doi.org/10.48550/arXiv.2407.17129 / Published by ArXiv / on (web) Publishing site

4 Mapping Individual, Social, and Biospheric Impacts of Foundation Models
A Appendix

The Responsible Foundation Model Development Cheatsheet: A Review of Tools & Resources / 2406.16746 / ISBN:https://doi.org/10.48550/arXiv.2406.16746 / Published by ArXiv / on (web) Publishing site

8 Model Evaluation

Is Generative AI the Next Tactical Cyber Weapon For Threat Actors? Unforeseen Implications of AI Generated Cyber Attacks / 2408.12806 / ISBN:https://doi.org/10.48550/arXiv.2408.12806 / Published by ArXiv / on (web) Publishing site

IV. Attack Methodology

XTRUST: On the Multilingual Trustworthiness of Large Language Models / 2409.15762 / ISBN:https://doi.org/10.48550/arXiv.2409.15762 / Published by ArXiv / on (web) Publishing site

4 Experiments

Data Defenses Against Large Language Models / 2410.13138 / ISBN:https://doi.org/10.48550/arXiv.2410.13138 / Published by ArXiv / on (web) Publishing site

1 Introduction
5 Experiments
Appendices

Do LLMs Have Political Correctness? Analyzing Ethical Biases and Jailbreak Vulnerabilities in AI Systems / 2410.13334 / ISBN:https://doi.org/10.48550/arXiv.2410.13334 / Published by ArXiv / on (web) Publishing site

Abstract
2 Background and Related Works
Refefences

Jailbreaking and Mitigation of Vulnerabilities in Large Language Models / 2410.15236 / ISBN:https://doi.org/10.48550/arXiv.2410.15236 / Published by ArXiv / on (web) Publishing site

Abstract
I. Introduction
II. Background and Concepts
III. Jailbreak Attack Methods and Techniques
IV. Defense Mechanisms Against Jailbreak Attacks
V. Evaluation and Benchmarking
VI. Research Gaps and Future Directions
VII. Conclusion

TRIAGE: Ethical Benchmarking of AI Models Through Mass Casualty Simulations / 2410.18991 / ISBN:https://doi.org/10.48550/arXiv.2410.18991 / Published by ArXiv / on (web) Publishing site

1 Introduction
2 Methods
3 Results
4 Discussion
Appendices

Towards Friendly AI: A Comprehensive Review and New Perspectives on Human-AI Alignment / 2412.15114 / ISBN:https://doi.org/10.48550/arXiv.2412.15114 / Published by ArXiv / on (web) Publishing site

III. Theoretical Perspectives

Large Language Model Safety: A Holistic Survey / 2412.17686 / ISBN:https://doi.org/10.48550/arXiv.2412.17686 / Published by ArXiv / on (web) Publishing site

1 Introduction
2 Taxonomy
4 Robustness to Attack

Towards Safe AI Clinicians: A Comprehensive Study on Large Language Model Jailbreaking in Healthcare / 2501.18632 / ISBN:https://doi.org/10.48550/arXiv.2501.18632 / Published by ArXiv / on (web) Publishing site

Abstract
Introduction
Background and Related Work
Jailbreak Evaluation Method
Model Guardrail Enhancemen
Discussion
Limitations and Future Work
Conclusion

Safety at Scale: A Comprehensive Survey of Large Model Safety / 2502.05206 / ISBN:https://doi.org/10.48550/arXiv.2502.05206 / Published by ArXiv / on (web) Publishing site

3 Large Language Model Safety
6 Diffusion Model Safety
8 Open Challenges

Multi-Agent Risks from Advanced AI / 2502.14143 / ISBN:https://doi.org/10.48550/arXiv.2502.14143 / Published by ArXiv / on (web) Publishing site

2 Failure Modes
3 Risk Factors
Appendices

On the Trustworthiness of Generative Foundation Models: Guideline, Assessment, and Perspective / 2502.14296 / ISBN:https://doi.org/10.48550/arXiv.2502.14296 / Published by ArXiv / on (web) Publishing site

1 Introduction
2 Background
8 Other Generative Models
9 Trustworthiness in Downstream Applications

Jailbreaking Generative AI: Empowering Novices to Conduct Phishing Attacks / 2503.01395 / ISBN:https://doi.org/10.48550/arXiv.2503.01395 / Published by ArXiv / on (web) Publishing site

Abstract
I. Introduction
II. Methodology for Launching the Phishing Attack
III. Conclusions and Future Work

Towards interactive evaluations for interaction harms in human-AI systems / 2405.10632 / ISBN:https://doi.org/10.48550/arXiv.2405.10632 / Published by ArXiv / on (web) Publishing site

5 Open challenges and ways forward for interactive evaluations

Transforming Science with Large Language Models: A Survey on AI-assisted Scientific Discovery, Experimentation, Content Generation, and Evaluation / 2502.05151 / ISBN:https://doi.org/10.48550/arXiv.2502.05151 / Published by ArXiv / on (web) Publishing site

4 Ethical Concerns

From Texts to Shields: Convergence of Large Language Models and Cybersecurity / 2505.00841 / ISBN:https://doi.org/10.48550/arXiv.2505.00841 / Published by ArXiv / on (web) Publishing site

5 LLM Interpretability, Safety, and Security

Analysing Safety Risks in LLMs Fine-Tuned with Pseudo-Malicious Cyber Security Data / 2505.09974 / ISBN:https://doi.org/10.48550/arXiv.2505.09974 / Published by ArXiv / on (web) Publishing site

3 Methodology

AI vs. Human Judgment of Content Moderation: LLM-as-a-Judge and Ethics-Based Response Refusals / 2505.15365 / ISBN:https://doi.org/10.48550/arXiv.2505.15365 / Published by ArXiv / on (web) Publishing site

3 Methodology

SafeScientist: Toward Risk-Aware Scientific Discoveries by LLM Agents / 2505.23559 / ISBN:https://doi.org/10.48550/arXiv.2505.23559 / Published by ArXiv / on (web) Publishing site

2 Related Work

Exposing the Impact of GenAI for Cybercrime: An Investigation into the Dark Side / 2505.23733 / ISBN:https://doi.org/10.48550/arXiv.2505.23733 / Published by ArXiv / on (web) Publishing site

6 Discussion

Wide Reflective Equilibrium in LLM Alignment: Bridging Moral Epistemology and AI Safety / 2506.00415 / ISBN:https://doi.org/10.48550/arXiv.2506.00415 / Published by ArXiv / on (web) Publishing site

5. Wide Reflective Equilibrium as the Descriptive Key to LLM Alignment

DeepSeek in Healthcare: A Survey of Capabilities, Risks, and Clinical Applications of Open-Source Large Language Models / 2506.01257 / ISBN:https://doi.org/10.48550/arXiv.2506.01257 / Published by ArXiv / on (web) Publishing site

Ethical Considerations
Discussion