Research and product · Security research

Nicholas Carlini

Known for breaking machine learning defenses. He won an obfuscated C contest with tic-tac-toe in a single printf call.

Role
Research Scientist
Joined Anthropic
2025
Before Anthropic
Google DeepMind

Career

  1. Anthropic
    • Research Scientist, safety and securityMar 2025 – present
  2. Google DeepMind
    • research scientist2023–2025
  3. Google Brain
    • research scientist2018–2023

Education

  • B.A., computer science and mathematics, UC Berkeley, 2013
  • Ph.D., Computer Science, UC Berkeley, 2018

Lesser-known facts

  • Built a working CPU inside Conway's Game of Life.
  • Spent seven years at Google Brain and DeepMind before joining in 2025.
  • Left Google DeepMind in March 2025 over disagreements about openly publishing security and privacy research, joining Anthropic "for a year."
  • Weighed founding a security nonprofit or taking a faculty job, and spent over 20 hours talking to Anthropic staff before deciding.
  • Used Claude Code to find remotely exploitable Linux kernel vulnerabilities, including one hidden for about 23 years.
  • Co-authored a poisoning attack that bought expired domains behind LAION-400M links, controlling about 0.01% of popular datasets.
  • Showed hidden voice commands could be embedded in audio to fool Mozilla's DeepSpeech, work covered by the New York Times in 2018.
  • Built a Doom clone in 13kB of JavaScript and a chess engine written in regular expressions.

Known for

  • Carlini & Wagner attack (2016)
  • "Towards Evaluating the Robustness of Neural Networks" (2017)
  • "Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examples" (2018)
  • "Poisoning the Unlabeled Dataset of Semi-Supervised Learning" (2021)
  • "Tight Auditing of Differentially Private Machine Learning" (2023)
  • "Stealing Part of a Production Language Model" (2024)

Honors

Best Student Paper Award, IEEE S&P 2017 · Best Paper Award, ICML 2018 · Distinguished Paper Award, USENIX 2021 · Distinguished Paper Award, USENIX 2023 · Best Paper Award, ICML 2024 (x2) · Best of Show, IOCCC 2020

Interviews and talks

Updated 2026-10-18 · 4 sources cited · Researched by Autumn · Research anyone like this

More research and product

All 42 people at Anthropic →