Apollo Research

Apollo Research, structured as a public benefit corporation, conducts AI safety research focused on detecting and mitigating ‘scheming’ — covert pursuit of misaligned goals — in frontier AI models, running pre-deployment evaluations for AI labs including Anthropic, Google DeepMind, Meta, Microsoft, and OpenAI, and building Watcher, a monitoring tool for coding-agent risks. It also advises governments on technical AI governance.