# A Peek Behind the Curtain: Using Step-Around Prompt Engineering to Identify Bias and Misinformation in GenAI Models arXiv: https://arxiv.org/abs/2503.15205 ยท HTML: https://mlsanity.com/paper/2503.15205/ ## Key facts - No citing papers in the map yet (data through 14 Oct 2026). - Placed in Safety alignment, refusal and guardrails, one of 12 directions in LLM safety and security. - Builds on 38 of its 75 references that are in the map. - First posted on arXiv 19 Mar 2025. ## Placed in - LLM safety and security / [Safety alignment, refusal and guardrails](https://mlsanity.com/safety/safety-alignment-refusal-and-guardrails/) ## Builds on (most cited first, 30 of 38) - [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020) - [Constitutional AI: Harmlessness from AI Feedback](https://mlsanity.com/paper/2212.08073/) (2022) - [Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing](https://mlsanity.com/paper/2107.13586/) (2021) - [Red Teaming Language Models to Reduce Harms: Methods, Scaling Behaviors, and Lessons Learned](https://mlsanity.com/paper/2209.07858/) (2022) - [A Prompt Pattern Catalog to Enhance Prompt Engineering with ChatGPT](https://mlsanity.com/paper/2302.11382/) (2023) - [Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpus](https://mlsanity.com/paper/2104.08758/) (2021) - [Certifying LLM Safety against Adversarial Prompting](https://mlsanity.com/paper/2309.02705/) (2023) - [PromptRobust: Towards Evaluating the Robustness of Large Language Models on Adversarial Prompts](https://mlsanity.com/paper/2306.04528/) (2023) - [A Comprehensive Survey of AI-Generated Content (AIGC): A History of Generative AI from GAN to ChatGPT](https://mlsanity.com/paper/2303.04226/) (2023) - [Large Language Models as General Pattern Machines](https://mlsanity.com/paper/2307.04721/) (2023) - [Constitutional Classifiers: Defending against Universal Jailbreaks across Thousands of Hours of Red Teaming](https://mlsanity.com/paper/2501.18837/) (2025) - [GPT detectors are biased against non-native English writers](https://mlsanity.com/paper/2304.02819/) (2023) - [The Malicious Use of Artificial Intelligence: Forecasting, Prevention, and Mitigation](https://mlsanity.com/paper/1802.07228/) (2018) - [Collective Constitutional AI: Aligning a Language Model with Public Input](https://mlsanity.com/paper/2406.07814/) (2024) - [The Prompt Report: A Systematic Survey of Prompt Engineering Techniques](https://mlsanity.com/paper/2406.06608/) (2024) - [Testing of Detection Tools for AI-Generated Text](https://mlsanity.com/paper/2306.15666/) (2023) - [Fairness And Bias in Artificial Intelligence: A Brief Survey of Sources, Impacts, And Mitigation Strategies](https://mlsanity.com/paper/2304.07683/) (2023) - [Red-Teaming for Generative AI: Silver Bullet or Security Theater?](https://mlsanity.com/paper/2401.15897/) (2024) - [Specific versus General Principles for Constitutional AI](https://mlsanity.com/paper/2310.13798/) (2023) - [GenAI Against Humanity: Nefarious Applications of Generative Artificial Intelligence and Large Language Models](https://mlsanity.com/paper/2310.00737/) (2023) - [Inverse Constitutional AI: Compressing Preferences into Principles](https://mlsanity.com/paper/2406.06560/) (2024) - [Diversity of Thought Improves Reasoning Abilities of LLMs](https://mlsanity.com/paper/2310.07088/) (2023) - [Disinformation Capabilities of Large Language Models](https://mlsanity.com/paper/2311.08838/) (2023) - [Lessons From Red Teaming 100 Generative AI Products](https://mlsanity.com/paper/2501.07238/) (2025) - [To Rely or Not to Rely? Evaluating Interventions for Appropriate Reliance on Large Language Models](https://mlsanity.com/paper/2412.15584/) (2024) - [Protecting Society from AI Misuse: When are Restrictions on Capabilities Warranted?](https://mlsanity.com/paper/2303.09377/) (2023) - [Game of Tones: Faculty detection of GPT-4 generated content in university assessments](https://mlsanity.com/paper/2305.18081/) (2023) - [Attack Atlas: A Practitioner's Perspective on Challenges and Pitfalls in Red Teaming GenAI](https://mlsanity.com/paper/2409.15398/) (2024) - [AI-generated faces influence gender stereotypes and racial homogenization](https://mlsanity.com/paper/2402.01002/) (2024) - [GenAI Detection Tools, Adversarial Techniques and Implications for Inclusivity in Higher Education](https://mlsanity.com/paper/2403.19148/) (2024) ## Abstract (arXiv) Step-around prompting is a form of adversarial prompt engineering in which a user strategically reframes, sequences, or contextualises requests to test whether a generative AI model's safety guardrails, alignment mechanisms, or bias mitigations can be undermined, inconsistently applied, or bypassed outright. This study examines this technique through the lens of academic ethics, situating it as a tool that has a clear impact on academic integrity, responsible conduct of research, duty of care to students, and institutional oversight of GenAI use in higher education. We argue that step-around prompting is one tool within the wider practice of audit, red-teaming, and institutional evaluation, and that its main value lies in documenting how representational, cultural, linguistic, disciplinary, and misinformation-related biases may appear across student-facing and research-facing uses of GenAI. To show why the ethical governance of this practice is required, we provide two illustrative examples of the technique in action, demonstrating how easily guardrails can be circumvented and what is at stake when they are. We clarify which bias categories are in scope and identify who should use the method and for what purposes. We conclude with an operational ethics-and-governance framework for controlled academic application, organised as two pillars (technical safeguards and ethical governance) and enacted through a decision and audit cycle that scales oversight to potential risk, grounded in harm minimisation, duty of care, transparency, proportionality, responsible disclosure, legal and contractual compliance, and student protection.