OpenAI · Thursday, October 1, 2026
OpenAI says it broke up a campaign to steal model reasoning
The File desk · Oct 1, 2026, 10:17 AM UTC
Status
Confirmed (OpenAI, Sept. 30, 2026). Moonshot’s reply is not in the post.
On Sept. 30, 2026, OpenAI said it identified and disrupted a coordinated campaign to extract “protected reasoning” — the model’s internal scratch work — without breaking encryption or stealing a database. Activity began around July 1. It spiked July 24–25, with about 16,000 extraction-pattern requests from more than 4,000 users. A related cluster of more than 15,000 users was disrupted by July 28. OpenAI attributes a core cluster to people associated with Moonshot AI, the developer of Kimi. It shared findings through the Frontier Model Forum.
On Sept. 30, 2026, OpenAI said it identified and disrupted a coordinated campaign to extract what it calls protected reasoning.
OpenAI describes that as the model’s internal scratch work.
The company says the campaign did not break encryption and did not steal a database.
OpenAI says the activity began around July 1.
It spiked on July 24 and July 25, with about 16,000 extraction-pattern requests from more than 4,000 users.
A related cluster of more than 15,000 users was disrupted by July 28.
OpenAI attributes a core cluster to people associated with Moonshot AI, the developer of Kimi.
OpenAI says it shared the findings through the Frontier Model Forum.
What is still unknown or disputed
- Moonshot’s reply.
- How much capability was successfully distilled.
Primary sources
Every claim in this story is drawn from the documents below. If a fetch failed, that is recorded on the card.
Source 1
Disrupting a coordinated model distillation campaign
OpenAI · September 30, 2026
OpenAI said it disrupted a campaign to extract protected reasoning without breaking encryption or stealing a database. Activity from about July 1, a July 24–25 spike of about 16,000 extraction-pattern requests from more than 4,000 users, and a related cluster of more than 15,000 users disrupted by July 28. A core cluster attributed to people associated with Moonshot AI. Findings shared through the Frontier Model Forum.
https://openai.com/index/disrupting-a-coordinated-model-distillation-campaign/