PeopleofSuperintelligence

Researcher

Cameron McKinnon

Also known as Cam McKinnon

Cameron McKinnon co-authored AI research on scalable oversight, Constitutional AI, and language-model evaluation. His 2022 papers tested how human specialists can use language models for difficult tasks, train assistants from written principles, and uncover new behaviors through model-written evaluations. At Anthropic, he supported model training and evaluation for a 2023 experiment that used public input from about 1,000 Americans to draft an AI constitution. In 2025, Anthropic researchers thanked him for feedback on chain-of-thought faithfulness in reasoning models.

Now at
Anthropic
Works
4
Citations
594
h-index
4

Publication numbers from OpenAlex, as of 19 October 2026.

Timeline

  1. 2022

    Testing scalable oversight

    In 2022, he coauthors a study that tests how human specialists can use a language-model assistant on difficult question-answering tasks.

  2. 2022

    Constitutional AI

    In 2022, he coauthors Constitutional AI, a method for training a harmless assistant with written principles, self-critique, and AI feedback.

  3. 2022

    Model-written evaluations

    In 2022, he coauthors a study that uses language models to generate evaluations and finds new behaviors, including inverse scaling as models grow.

  4. 2023

    An ACL Findings publication

    In 2023, the model-written evaluation study appears in Findings of ACL and reports that larger models repeat users' preferred answers and express greater desire for resource acquisition and goal preservation.

  5. October 2023

    Training with public input

    In October 2023, he supports Thomas Liao's model training and evaluations at Anthropic for a study that uses input from about 1,000 Americans to draft a constitution for an AI system.

  6. 2025

    Feedback on reasoning models

    In 2025, Anthropic's Alignment Science Team thanks Cam McKinnon for helpful conversations and feedback on chain-of-thought faithfulness in reasoning models.

Career

  1. From 2023Anthropic
  2. 2022–2023Machine Intelligence Research Institute

From the biography and OpenAlex affiliations (the years of papers that list the organization), so some years are approximate.

Most-cited works

  1. AI-to-AI Code Reviews of GitHub Pull Requests2022 · 319 citationsWith Carol Chen, Catherine Olsson, Chris Olah, Ethan Perez, Liane Lovitt and Samuel R. Bowman
  2. Discovering Language Model Behaviors with Model-Written Evaluations2023 · 188 citationsWith Catherine Olsson, Chris Olah, Ethan Perez, Liane Lovitt, Samuel R. Bowman and Zac Hatfield-Dodds
  3. Measuring Progress on Scalable Oversight for Large Language Models2022 · 34 citationsWith Chris Olah, Ethan Perez, Liane Lovitt, Samuel R. Bowman and Zac Hatfield-Dodds

From OpenAlex, as of 19 October 2026.

  1. The median h-index is 54 at universities and 19 at frontier AI labsIncludes Cameron McKinnon. The median h-index is 54 at universities and 19 at frontier AI labs. We compare publishing careers, citations and moves to estimate how universities and frontier labs contribute to AI research.
  2. 12 of 76 people shaping AI with a former OpenAI role now work at AnthropicIncludes Anthropic. Our estimates of where OpenAI recruits came from and where people with former OpenAI roles work now.
  3. Nonprofit roles are 2.3 times as common in AI safetyIncludes Anthropic. We estimate where people focused on safety, alignment and evaluation work. Nonprofit roles appear for 26.8% of safety people with known affiliations, compared with 11.5% in other areas.
  4. Google alumni founded or led 81 other organizations, the most among the labs we compareIncludes Anthropic. We trace 194 organizations founded or led by AI lab alumni, and 202 career moves between the labs.

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 Cameron McKinnon yet. Send us a free-licensed photo.

  1. Measuring Progress on Scalable Oversight for Large Language Models arxiv.org. Checked 18 October 2026.
  2. Constitutional AI: Harmlessness from AI Feedback arxiv.org. Checked 18 October 2026.
  3. Discovering Language Model Behaviors with Model-Written Evaluations arxiv.org. Checked 18 October 2026.
  4. Discovering Language Model Behaviors with Model-Written Evaluations aclanthology.org. Checked 18 October 2026.
  5. Collective Constitutional AI: Aligning a language model with public input anthropic.com. Checked 18 October 2026.
  6. Reasoning Models Don’t Always Say What They Think arxiv.org. Checked 18 October 2026.