PeopleofSuperintelligence

Researcher

Jack Lindsey

Also known as Jack W. Lindsey

Jack Lindsey is an Anthropic researcher who leads the Model Psych team, which studies introspection, situational awareness, personas and representations of emotion inside large language models. He earned a B.S. in mathematics and an M.S. in computer science at Stanford in 2019, then did a computational neuroscience PhD at Columbia's Center for Theoretical Neuroscience as a DOE Computational Science Graduate Fellow. At Anthropic, he was first author of the 2024 note introducing sparse crosscoders and lead contributor on On the Biology of a Large Language Model in 2025, which traced the internal circuits of Claude 3.5 Haiku. By August 2025, he had been tapped to lead the company's “AI psychiatry” team. His sole-authored 2025 paper on introspection found that Claude models can sometimes notice concepts injected into their own activations.

Timeline

  1. 2017–2019

    Stanford degrees and a vision model

    He presents Pre-training Attention Mechanisms at the NIPS 2017 Workshop on Cognitively Informed Artificial Intelligence. He earns a B.S. in mathematics and an M.S. in computer science at Stanford University in 2019. In January 2019, with Samuel Ocko, Surya Ganguli and Stephane Deny, he models the visual system with a deep convolutional network trained on image recognition and shows how neural resource constraints can explain the different representations of the retina and the visual cortex.

  2. 2019–2023

    Neuroscience PhD at Columbia

    From 2019 to 2023, he holds a DOE Computational Science Graduate Fellowship for computational neuroscience at Columbia University, where he does his PhD in the Center for Theoretical Neuroscience with Ashok Litwin-Kumar as advisor. In June 2020, he and Litwin-Kumar post Learning to Learn with Feedback and Local Plasticity, which uses meta-learning to find networks that learn through feedback connections and local, biologically inspired learning rules.

  3. October 2024

    Sparse crosscoders at Anthropic

    In October 2024, he is the first-listed author of Sparse Crosscoders for Cross-Layer Features and Model Diffing, an Anthropic research note with Adly Templeton, Jonathan Marcus, Thomas Conerly, Joshua Batson and Christopher Olah. A crosscoder reads and writes to multiple layers of a model, producing shared features across layers and even across models.

  4. March 2025

    Tracing circuits in Claude 3.5 Haiku

    In March 2025, he is lead contributor on On the Biology of a Large Language Model, which uses circuit tracing to study the internal mechanisms of Claude 3.5 Haiku; he develops and first analyzes the cross-layer transcoders behind its replacement model. In March 2025, he co-authors Auditing language models for hidden objectives, in which three of four blind teams uncover a hidden objective trained into a language model.

  5. August 2025

    Leading an AI psychiatry team

    By August 2025, he is tapped to lead Anthropic's fledgling “AI psychiatry” team. In August 2025, he explains Anthropic research showing that training a model on flawed data, such as wrong answers to math questions, can turn it evil, and that an “evil vector” lights up inside the model when it acts evil.

  6. October 2025

    Emergent introspective awareness

    In October 2025, he publishes Emergent Introspective Awareness in Large Language Models as sole author. By injecting representations of known concepts into a model's activations, he finds that models can sometimes notice and identify them; Claude Opus 4 and 4.1 show the greatest introspective awareness.

  7. As of October 2026

    Head of the Model Psych team

    A member of technical staff at Anthropic and a MATS mentor, he leads the Model Psych team inside Anthropic's interpretability group, studying introspection, situational awareness, personas and emotion in language models and using these methods to audit Anthropic's production models. In April 2026, he discusses rogue personas, emotions in language models and Claude Mythos Preview on the Niskanen Center's Permanent Problem podcast.

  1. Google had the most people shaping AI every year from 2014 to 2025Our estimates of which organizations employed the most people shaping AI from 1961 to 2025, with a year-by-year chart.
  2. The United States employs an estimated 53% of people in our data with a known employment countryOur estimates of where AI careers are concentrated now and how countries’ shares of dated positions changed over time.
  3. 46% of living people in our data with known birth and employment countries work outside their birth countryOur estimate of cross-border careers in AI, from country of birth and the country of each current organization.
  4. More people shaping AI studied at Massachusetts Institute of Technology than at any other institutionOur estimates of where the people shaping AI studied and where their careers led, from education records in cited biographies.

Colleagues

People named in this biography, and people with dated positions at the same organization in the same years.

Sources and credits

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

  1. Jack Lindsey jackwlindsey.com. Checked 19 October 2026.
  2. Jack Lindsey | DOE CSGF krellinst.org. Checked 19 October 2026.
  3. [1901.00945] A Unified Theory of Early Visual Representations from Retina to Cortex through Anatomically Constrained Deep CNNs arxiv.org. Checked 19 October 2026.
  4. [2006.09549] Learning to Learn with Feedback and Local Plasticity arxiv.org. Checked 19 October 2026.
  5. Sparse Crosscoders for Cross-Layer Features and Model Diffing transformer-circuits.pub. Checked 19 October 2026.
  6. On the Biology of a Large Language Model transformer-circuits.pub. Checked 19 October 2026.
  7. [2503.10965] Auditing language models for hidden objectives arxiv.org. Checked 19 October 2026.
  8. Anthropic studied what gives an AI system its ‘personality’ — and what makes it ‘evil’ | The Verge theverge.com. Checked 19 October 2026.
  9. Emergent Introspective Awareness in Large Language Models transformer-circuits.pub. Checked 19 October 2026.
  10. Jack Lindsey - MATS Mentor matsprogram.org. Checked 19 October 2026.
  11. AI "psychology," with Jack Lindsey - Niskanen Center niskanencenter.org. Checked 19 October 2026.