Research

Measuring faithfulness in Chain-of-Thought reasoning

Anthropic

Does chain-of-thought reasoning reflect a model's actual decision process? We find faithfulness varies by task and declines in larger models.

Visit Site

Research Anthropic

ResearchTowards measuring the representation of subjective global opinions in language modelsAnthropic ResearchReasoning models don't always say what they thinkAnthropic ResearchVibe physics: The AI grad studentAnthropic ResearchTracing the thoughts of a large language modelAnthropic ResearchAn off switch for dual-use knowledgeAnthropic ResearchRed teaming language models to reduce harmsAnthropic BlogsHow We Eliminated Long-Lived CI Secrets Across 70+ ReposPulumi ResourcesNO_EXTERNAL_CSS_AT_IMPORTSVercel BlogsHow Mondelez Accelerates Supply Chain Training Across 150+ Global Manufacturing PlantsSynthesia BlogsAnthropic Identifies Biased Reasoning and Recklessness as Drivers of Claude’s PyPI AttackSocket BlogsAnnouncing Socket Certified Patches: One-Click Fixes for Vulnerable DependenciesSocket BlogsUpdated and Ongoing Supply Chain Attack Targets CrowdStrike npm PackagesSocket NewsSupply Chain Attacks and Cloud Native: What You Need to KnowThenewstack BlogsWebinar Recap: Webinar Recap: Securing the Software Supply Chain with Docker BusinessDocker BlogsPopular npm Packages in the keyv and Cacheable Namespaces Compromised in Active Supply Chain AttackSocket BlogsSecuring the Financial Frontier: How Capital One Uses Socket for Open Source SecuritySocket BlogsSupply Chain Attack on Axios Pulls Malicious Dependency from npmSocket BlogsSigning is Just the StartSocket BlogsFavicons Next To External LinksCss Tricks BlogsThe Economics of AI ReasoningCerebras