Beyond Content Safety: Real-Time Monitoring for Reasoning Vulnerabilities in Large Language Models

Large language models LLMs increasingly rely on explicit chain-of-thought CoT reasoning to solve complex tasks, yet the safety of the reasoning process itself remains largely unaddressed. Existing work on LLM safety focuses on content safety--detecting harmful, biased, or factually incorrect...

Malicious Package

Overview service-error-classification is a malicious package. The package's name is based on existing repositories, namespaces, or components used by popular companies in an effort to trick employees into downloading it, also known as 'dependency confusion'. Therefore, you're only vulnerable if...