Skip to content
Longterm Wiki

Redwood Research

Safety Organization
Founded Jun 2021 (4 years old)HQ: San Francisco, CAredwoodresearch.org

Also known as: Redwood

DateEventTypeDescriptionSource
Sep 2021Tax-exempt status granted; 10 staff assembledFoundingprojects.propublica.org (opens in new tab)
Dec 2021MLAB bootcamp launchesLaunchInaugural ML for Alignment Bootcamp with 40 participants; 3-week intensive teaching attendees to build BERT/GPT-2 from scratch.blog.redwoodresearch.org (opens in new tab)
2022Adversarial robustness research projectMilestoneInitial adversarial training project; later acknowledged by leadership as unsuccessful.blog.redwoodresearch.org (opens in new tab)
2022Causal scrubbing methodology developedPublicationDeveloped across 2022-2023; method for rigorously testing mechanistic interpretability claims.lesswrong.com (opens in new tab)
2023REMIX interpretability program runsLaunchMechanistic interpretability training program for ~10-15 junior researchers.forum.effectivealtruism.org (opens in new tab)
2024Buck Shlegeris becomes CEO; AI Control ICML oralLeadership ChangeBuck Shlegeris transitions from CTO to CEO and Director; Ryan Greenblatt serves as Chief Scientist. AI Control work accepted as an ICML oral.projects.propublica.org (opens in new tab)
Dec 2024Alignment faking paper with AnthropicPublicationLandmark collaboration with Anthropic on alignment faking research.anthropic.com (opens in new tab)