Mindgard blog analyzes model extraction and distillation attacks

Cited in the reports

• Stealing Reasoning Traces from Proprietary LLM APIs

  “Panfilov et al. (August 2026) found that these encrypted blocks were interchangeable across models from the same provider.”

  What is a model extraction attack? How distillation attacks steal AI models through APIs

  Using cross-model replay of encrypted chain-of-thought blocks to induce a weak model to transcribe proprietary reasoning chains verbatim

  arXiv · 2026-08-10 · Deep reading

• Stealing Part of a Production Language Model

  “In Stealing Part of a Production Language Model (Carlini et al., 2024), researchers from Google DeepMind, ETH Zurich and partner institutions recovered the full embedding projection layer of OpenAI’s”

  What is a model extraction attack? How distillation attacks steal AI models through APIs

  Proposes a black-box model stealing attack that extracts the embedding projection layer of a production language model through API queries

  arXiv · 2024-03-11 · Deep reading

• Stealing Machine Learning Models via Prediction APIs

  “The 2016 paper that named the problem, Stealing Machine Learning Models via Prediction APIs, showed near-perfect fidelity copies of logistic regression, neural network and decision tree models hosted”

  What is a model extraction attack? How distillation attacks steal AI models through APIs

  Propose equation-solving and decision-tree path-search attacks to steal an equivalent model through a prediction API

  arXiv · 2016-09-09 · Deep reading