Skip to main content

    Ryan Greenblatt

    Technical Staff at METR who researches AI control, empirical risk evaluation, and defenses against subversive model behavior.

    #68 AI Security#64 AI Safety
    15CiteRank

    Last updated 11 Nov 2026 · Named in 3 answers across 3 engines
    Bing AI Mode starts a separate baseline. Copilot remains a separate public engine and keeps its history; changes across the switch are not like-for-like growth.

    How AI engines see them

    Recognized by 3 of 3 measured engines. Updated 2026-11-07

    The recognizing engines agree that Ryan Greenblatt is an AI safety and security researcher associated with Redwood Research whose work focuses on AI control, deceptive alignment, and adversarial red teaming. They emphasize his research on preventing models from strategically faking compliance or circumventing oversight. Chat_gpt adds his affiliation with METR and co-authorship of foundational papers on AI control. Google_ai_mode uniquely highlights his work on model organisms of misalignment. The most frequently cited domains are redwoodresearch.org, youtube.com, and alignmentproject.aisi.gov.uk.

    • ChatGPTNot named

      Ryan Greenblatt is an AI security and control researcher focusing on adversarial evaluation, deceptive alignment, and systems designed to withstand intentional subversion.

      Known for
      • AI control
      • adversarial evaluation
      • deceptive alignment research
    • GeminiNot named

      Ryan Greenblatt is an AI safety and security researcher who investigates AI control, deceptive alignment, model monitoring, and alignment faking in frontier models.

      Known for
      • AI Control paradigm
      • alignment faking studies
      • safeguarding untrusted systems
    • Google AI ModeNot named

      Ryan Greenblatt is an AI safety researcher focused on alignment faking, deceptive model behavior, model organisms of misalignment, and agent red-teaming pipelines.

      Known for
      • alignment faking research
      • model organisms of misalignment
      • agent and monitor red-teaming pipelines
    • PerplexityNot named

      No description yet.

    • Bing AI ModeNot named

      No description yet.

    • CopilotNot named

      No description yet.

    Top Segments

    • #68AI Securityof 86
      Named in 2 answers
    • #64AI Safetyof 78
      Named in 3 answers

    Track Ryan

    Follow Ryan's measured AI visibility and see the questions where they rank.