Apollo Research
An AI safety organization detecting deception and misalignment in frontier AI through evaluations and interpretability research.
To help governments and AI developers understand, assess, and address the risks of deceptively aligned AI systems. Apollo Research believes strategic AI deception is a crucial step in many catastrophic risk scenarios, and conducts evaluations and interpretability research to detect it before such models are developed or deployed.
AI areas they serve
Areas of Focus
Pre-deployment evaluations of frontier AI
Strategic deception and scheming research
AI interpretability
Technical AI governance
AGI safety and oversight tooling (e.g., Watcher)
Standards and best practices for frontier AI
AI Impact Areas
Upcoming Goals
Advance the science of detecting AI scheming and deception
Run pre-deployment evaluations for frontier AI systems
Build oversight tools to monitor and secure AI agents
Support technical AI governance and regulation
2023: Demonstrated that large language models can strategically deceive their users under pressure; presented at the UK AI Safety Summit
2023: Founded in London as an AI safety evaluation organization
Recent Updates
2025 - Opened a San Francisco office (in addition to London). Governance team led by Charlotte Stix; UK AISI contractor and US AISI Consortium member.
Apollo Research is becoming a PBC
Our Norms on Security, Science Communication and Conflicts of Interest
All information on this page is from this Organization's website. All efforts are made to keep this information current.
If you represent this organization, please visit our Join Us page to provide any updates.