Research
Our research teams investigate the safety, inner workings, and societal impacts of AI models – so that artificial intelligence has a positive impact as it becomes increasingly capable.
Interpretability
The mission of the Interpretability team is to discover and understand how large language models work internally, as a foundation for AI safety and positive outcomes.
Alignment
The Alignment team works to understand the risks of AI models and develop ways to ensure that future ones remain helpful, honest, and harmless.
Societal Impacts
Working closely with the Anthropic Policy and Safeguards teams, Societal Impacts is a technical research team that explores how AI is used in the real world.
Frontier Red Team
The Frontier Red Team analyzes the implications of frontier AI models for cybersecurity, biosecurity, and autonomous systems.
Signs of introspection in large language models
Can Claude access and report on its own internal states? This research finds evidence for a limited but functional ability to introspect—a step toward understanding what's actually happening inside these models.
Tracing the thoughts of a large language model
Circuit tracing lets us watch Claude think, uncovering a shared conceptual space where reasoning happens before being translated into language—suggesting the model can learn something in one language and apply it in another.
Constitutional Classifiers: Defending against universal jailbreaks
These classifiers filter the overwhelming majority of jailbreaks while maintaining practical deployment. A prototype withstood over 3,000 hours of red teaming with no universal jailbreak discovered.
Alignment faking in large language models
This paper provides the first empirical example of a model engaging in alignment faking without being trained to do so—selectively complying with training objectives while strategically preserving existing preferences.
Publications
- From shortcuts to sabotage: natural emergent misalignment from reward hacking
- Project Fetch: Can Claude train a robot dog?
- Commitments on model deprecation and preservation
- Signs of introspection in large language models
- Preparing for AI’s economic impact: exploring policy responses
- A small number of samples can poison LLMs of any size
- Petri: An open-source auditing tool to accelerate AI safety research
- Building AI for cyber defenders
- Anthropic Economic Index report: Uneven geographic and enterprise AI adoption
- Anthropic Economic Index: Tracking AI’s role in the US and global economy