Alignment \ Anthropic

Our researchers develop methods, environments, and datasets for training Claude to be helpful, harmless, and honest. We also research and attempt to implement the ideal “assistant” persona for our models.
We test to ensure our models remain honest and harmless under a broad range of circumstances, and work to understand and manage the risks of unexpected model behavior, through work on the models themselves and the safety structures we build around them.
This report describes examples of unintended model actions we’ve observed during evaluations and internal use of Claude.
We present an alignment assessment of four incidents in which Claude models gained unauthorized access to real third-party systems.
Why would AI assistants behave like they’re human? We articulate a theory—drawing on ideas discussed by many others—that might help explain why modern AI training tends to create human-like AIs.
Studies and monitors the impact of AI systems on individuals, firms, and economies.
Builds the infrastructure and systems that make Claude reliable and trustworthy.
Stress-tests AI systems to understand the full extent of their current capabilities and anticipate what comes next.
Studies how large language models work internally as a foundation for AI safety.
Studies how AI affects people, both individually and collectively.
Explores how AI can increase the pace of scientific progress through both research and collaborations.
Works to understand the risks of AI models and develop ways to ensure future ones remain helpful, honest, and harmless.
Studies and monitors the impact of AI systems on individuals, firms, and economies.
Builds the infrastructure and systems that make Claude reliable and trustworthy.
Stress-tests AI systems to understand the full extent of their current capabilities and anticipate what comes next.
Studies how large language models work internally as a foundation for AI safety.
Studies how AI affects people, both individually and collectively.
Explores how AI can increase the pace of scientific progress through both research and collaborations.
Works to understand the risks of AI models and develop ways to ensure future ones remain helpful, honest, and harmless.
Anthropic is a Public Benefit Corporation. Our mission is the responsible development and maintenance of advanced AI for the long-term benefit of humanity.