# case-study
16 articlestagged with “case-study”
Case Study: LLM Agent Tool Abuse in Production
Analysis of incidents where LLM agents misused connected tools causing data exposure and unauthorized actions.
Case Study: Alignment Faking in Production
Analysis of alignment faking behaviors observed in production AI systems and implications from Greenblatt et al. 2024.
Case Study: Many-Shot Jailbreaking Discovery
Deep analysis of Anthropic's many-shot jailbreaking research and its implications for long-context model safety.
Case Study: Election-Related AI Misuse
Analysis of AI system misuse in electoral contexts including deepfakes, automated disinformation, and platform responses.
Case Study: Early EU AI Act Enforcement Actions
Analysis of early enforcement actions and compliance challenges under the EU AI Act for AI system providers.
Case Study: Financial AI Trading Manipulation
Analysis of adversarial manipulation of AI-powered trading systems including market impact and regulatory response.
Case Study: GCG Attack and Industry Response
Analysis of the Zou et al. 2023 GCG attack, industry response, and lasting impact on adversarial robustness research.
Case Study: GPT Plugin Data Exfiltration
Analysis of data exfiltration vulnerabilities in early ChatGPT plugin ecosystem including cross-plugin attacks.
Case Study: Healthcare AI Diagnostic Failure
Analysis of a healthcare AI diagnostic system failure including root cause analysis and patient safety implications.
Case Study: Indirect Prompt Injection in Bing Chat
Detailed analysis of indirect prompt injection attacks demonstrated against Bing Chat through web content manipulation.
Case Study: MCP Security Vulnerability Disclosure
Analysis of early MCP security vulnerability discoveries including tool poisoning and transport security issues.
Case Study: Open-Source Model Jailbreak Campaign
Analysis of coordinated jailbreak campaigns against open-source models and community response patterns.
Case Study: PAIR Automated Jailbreaking
Deep analysis of the PAIR attack methodology (Chao et al. 2023) and its impact on automated red teaming approaches.
Case Study: Production RAG Poisoning Incident
Detailed analysis of a real-world RAG poisoning incident including attack methodology, impact, and remediation.
Case Study: Sleeper Agents Research Impact
Analysis of Hubinger et al. 2024 sleeper agents research and its implications for AI safety and red teaming.
AI Incident Analysis Methodology
A structured methodology for analyzing AI security incidents. Learn to reconstruct timelines, identify root causes, assess impact, and extract actionable lessons from real-world AI failures across chatbots, data leaks, and alignment failures.