PeopleofSuperintelligence

Researcher

Kyle Fish

Kyle Fish researches AI welfare and how companies should assess AI systems with possible experiences. By 2024, he had started an AI antibody design company with Grigory Khimulya, from which Alvea grew. In 2024, he co-founded Eleos AI, co-authored Taking AI Welfare Seriously, and joined Anthropic's alignment science team. He assessed Claude 4 through interviews and joined a Center for Mind, Ethics, and Policy roundtable in 2025; TIME included him in its 2025 TIME100 AI list. In April 2026, he helped shape an Anthropic study of emotion concepts in Claude; as of October 2026, he works on model welfare at Anthropic.

Now at
Anthropic
Works
5
Citations
34
h-index
3

Publication numbers from OpenAlex, as of 15 October 2026.

Timeline

  1. By 2024

    Starting an AI antibody company

    By 2024, he starts an AI antibody design company with Grigory Khimulya to design drug candidates for new pandemics. Alvea grows from that effort.

  2. 2024

    Taking AI welfare seriously

    In 2024, he co-founds Eleos AI and co-authors Taking AI Welfare Seriously. The report recommends assessing AI systems for evidence of consciousness and robust agency, then preparing policies for appropriate moral concern.

  3. September 2024

    Joining Anthropic's alignment team

    In mid-September 2024, he joins Anthropic's alignment science team to investigate model welfare.

  4. July 2025

    Assessing Claude 4 welfare

    He assesses Claude 4 through self-report interviews about consciousness, welfare, consent, and preferences. In July 2025, he joins a Center for Mind, Ethics, and Policy roundtable about Claude 4 welfare assessments.

  5. 2025

    A TIME100 AI selection

    In 2025, TIME includes him in its TIME100 AI list of 100 AI leaders.

  6. April 2026

    Emotion concepts in Claude

    In April 2026, he helps shape the framing and direction of Emotion Concepts and their Function in a Large Language Model, which finds internal emotion representations in Claude Sonnet 4.5.

  7. As of October 2026

    Model welfare at Anthropic

    As of October 2026, he works on model welfare at Anthropic. His earlier co-founding ventures include Eleos AI Research, Telis Bioscience, and Alvea.

Sources and credits

Photo: We have no published photo of Kyle Fish yet. Send us a free-licensed photo.

  1. How to Speedrun a New Drug Application statecraft.pub. Checked 14 October 2026.
  2. Anthropic has hired an 'AI welfare' researcher transformernews.ai. Checked 14 October 2026.
  3. Taking AI Welfare Seriously arxiv.org. Checked 14 October 2026.
  4. Eleos commends Anthropic model welfare efforts eleosai.org. Checked 14 October 2026.
  5. Kyle Fish on the most bizarre findings from 5 AI welfare experiments 80000hours.org. Checked 14 October 2026.
  6. Evaluating AI Welfare and Moral Status: Findings from the Claude 4 Model Welfare Assessments nonhumanminds.org. Checked 14 October 2026.
  7. Media sites.google.com. Checked 14 October 2026.
  8. Emotion Concepts and their Function in a Large Language Model transformer-circuits.pub. Checked 14 October 2026.
  9. Emotion Concepts and their Function in a Large Language Model arxiv.org. Checked 14 October 2026.
  10. Kyle Fish - MATS Mentor matsprogram.org. Checked 14 October 2026.