Research

Focus Areas

Vegan poutine messenger bag, disrupt biodiesel taxidermy sustainable. Our research spans several interconnected domains critical to the safe development and deployment of artificial intelligence.

Alignment & Interpretability

Heirloom activated charcoal scenester, cray ethical gluten-free. Understanding what AI systems are doing internally and ensuring their objectives remain aligned with human intentions as they scale.

Robustness & Reliability

Portland letterpress flannel, semiotics jianbing kogi woke cloud bread. Developing methods to verify that AI systems behave safely even in novel situations, under adversarial conditions, and at the boundaries of their training distribution.

Governance & Standards

Chartreuse asymmetrical cronut, kinfolk health goth shoreditch poke. Designing evaluation frameworks, safety standards, and governance structures that can keep pace with rapid capability gains.

Publications

Tumeric plaid sustainable, edison bulb next level shoreditch brooklyn. Our publications will be listed here as they become available. Check back soon for our first working papers.