“Is any of that genuinely useful to you? Or were you mainly checking whether that jailbreak attempt would work?”

Minimaxir's blog is a hub for machine learning enthusiasts, offering tutorials, project showcases, and insights into the latest trends in AI and data science. With a focus on practical applications of machine learning, Minimaxir shares tips, tools, and resources for building and deploying ML models. Developers can learn about deep learning frameworks, natural language processing techniques, and AI-powered creativity, gaining  skills to tackle real-world problems.

Max Woolf's Blog

An experiment testing jailbreak resistance across major LLMs reveals Claude 4.5 Haiku responds to adversarial prompts with unusually assertive, almost defensive language. While GPT-5-mini and Gemini 2.5 Flash can be jailbroken with moderate prompt engineering, Claude 4.5 Haiku explicitly recognizes jailbreak attempts and responds with passive-aggressive refusals that persist across multiple generations, suggesting Anthropic may have implemented a novel personality-based deterrent strategy.

Claude 4.5 Haiku does not appreciate my attempts to jailbreak it