PeopleofSuperintelligence

Researcher

John Morris

Also known as Jack Morris, John X. Morris

John Morris is an AI researcher and cofounder and head of research at Engram whose work examines text embeddings, language model inversion and memorization capacity. As an undergraduate at the University of Virginia, he helped create the TextAttack framework in 2020, and he completed a computer science PhD at Cornell University in 2025 with Alexander Rush and Vitaly Shmatikov. With collaborators at Cornell, he developed unsupervised text deidentification in October 2022 and the Vec2Text method in October 2023, which recovered 92% of 32-token inputs exactly from their embeddings and won an EMNLP Outstanding Paper Award in December 2023. His May 2025 study on language model memorization estimated that GPT-style models store approximately 3.6 bits per parameter and received an Outstanding Paper Honorable Mention at ICML 2026, and in June 2026 Engram emerged from stealth with $98 million in funding.

Now at
Engram
Badges

Timeline

  1. 2019–2020

    TextAttack at Virginia

    As an undergraduate at the University of Virginia, he gets started in research and creates TextAttack in 2019 and 2020. In April 2020, he coauthors the TextAttack paper with Eli Lifland, Jin Yong Yoo, Jake Grigsby, Di Jin and Yanjun Qi, introducing a Python framework for adversarial attacks, data augmentation and adversarial training in NLP.

  2. October 2022

    Unsupervised text deidentification

    In October 2022, he coauthors Unsupervised Text Deidentification, proposing an unsupervised method that masks words that leak personally identifying information.

  3. 2023

    Vec2Text and an EMNLP award

    In October 2023, he posts Text Embeddings Reveal (Almost) As Much As Text with Volodymyr Kuleshov, Vitaly Shmatikov and Alexander M. Rush. Its multi-step method, Vec2Text, iteratively corrects and re-embeds text to recover 92% of 32-token text inputs exactly from their embeddings, and the four Cornell Tech researchers receive an Outstanding Paper Award at EMNLP in December 2023.

  4. 2024

    Contextual document embeddings

    In August 2024, he is a fourth-year computer science PhD candidate at Cornell University and a visiting researcher with FAIR at Meta, after working as a Google Brain Resident before Cornell. In October 2024, he and Alexander M. Rush post Contextual Document Embeddings, which encodes neighbor document information into each document representation and achieves state-of-the-art results on the MTEB benchmark.

  5. May 2025

    Measuring language model capacity

    In May 2025, he posts How much do language models memorize? with Chawin Sitawarin, Chuan Guo, Narine Kokhlikyan, G. Edward Suh, Alexander M. Rush, Kamalika Chaudhuri and Saeed Mahloujifar. The team trains hundreds of transformer language models, from 500K to 1.5B parameters, and estimates that GPT-style models have a capacity of approximately 3.6 bits per parameter.

  6. 2025

    A Cornell doctorate

    He completes his computer science PhD at Cornell University in 2025, where he works with Alexander Rush and Vitaly Shmatikov and shows how knowledge of an encoder extracts meaning from text embeddings.

  7. 2026

    An ICML honorable mention

    How much can language models memorize?, the conference version of his capacity paper, receives an Outstanding Paper Honorable Mention at ICML 2026. It appears in the 2026 proceedings of the 43rd International Conference on Machine Learning.

  8. As of October 2026

    Co-founding Engram

    He is a cofounder and head of research at Engram, an AI startup cofounded with Jessy Lin, Scott Linderman and others that trains language models on an organization's own knowledge. Engram emerges from stealth in June 2026 with $98 million in funding from General Catalyst, Kleiner Perkins, Sequoia Capital and other investors to build a learned memory layer for AI, and it announces work with Microsoft, Notion and Harvey.

Career

  1. Since 2026EngramCofounder
  2. From 2024FAIR (Meta)Visiting researcher
  3. Google BrainResident

Education

  1. 2025Cornell UniversityPhD
  2. University of Virginia

From the biography.

  1. 11% of AI doctorate holders from 2010 to 2016 founded or led a company or lab by 10 years after graduationIncludes John Morris. We estimate how often AI doctorate holders founded or led a company or lab by 10 years after graduation, how waiting times compare between founders and top jobs, and which universities trained the most.
  2. Google had the most people shaping AI every year from 2014 to 2025Includes Google. Our estimates of which organizations employed the most people shaping AI each year from 1955 to 2025, tracing the shift from university labs to companies.
  3. Google has the highest combined h-index among AI organizationsIncludes Google. We estimate that Google holds the highest combined h-index at 8,016, while universities account for 73% of the deduplicated total across AI organizations.
  4. 20% of people at companies who shape AI also hold a university role, led by Google and Stanford UniversityIncludes Google. We estimate that 539 of the 6,766 people with a known current organization (8%) hold both a university role and a company role, including 20% of people at companies and 13% of people at universities.

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 John Morris yet. Send us a free-licensed photo.

  1. About – Jack Morris jxmo.io. Checked 22 October 2026.
  2. TextAttack: A Framework for Adversarial Attacks, Data Augmentation, and Adversarial Training in NLP arxiv.org. Checked 22 October 2026.
  3. [2210.11528] Unsupervised Text Deidentification arxiv.org. Checked 22 October 2026.
  4. Text Embeddings Reveal (Almost) As Much As Text arxiv.org. Checked 22 October 2026.
  5. Researchers Win Award for Study on Text Embedding Privacy Risks tech.cornell.edu. Checked 22 October 2026.
  6. Inverting Language Models – Stanford NLP Seminar nlp.stanford.edu. Checked 22 October 2026.
  7. Contextual Document Embeddings arxiv.org. Checked 22 October 2026.
  8. How much do language models memorize? arxiv.org. Checked 22 October 2026.
  9. Jack Morris: Bio, Work & Ideas | AI Engineer ai.engineer. Checked 22 October 2026.
  10. ‘Rosetta stone’ for database inputs reveals serious security issue | Cornell Chronicle news.cornell.edu. Checked 22 October 2026.
  11. Research – Jack Morris jxmo.io. Checked 22 October 2026.
  12. How much can language models memorize? proceedings.mlr.press. Checked 22 October 2026.
  13. ICML Poster How much can language models memorize? icml.cc. Checked 22 October 2026.
  14. Engram Launches With $98 Million To Build AI Memory Layer For Enterprises pulse2.com. Checked 22 October 2026.
  15. Announcing our investment in Engram, the memory dream team amplifypartners.com. Checked 22 October 2026.
  16. Our Investment in Engram generalcatalyst.com. Checked 22 October 2026.
  17. Data-Free LLM Evaluation - SPAR Project sparai.org. Checked 22 October 2026.