Skip to content
NewA calmer, faster ZeroHour
AI safety & securityhighExploited in the wild

OpenAI Blocks 15,000 Requests Trying to Extract Protected Model Reasoning

GBHackersby Divya ()2 min read3 · 1 read
AI summary · grok-4.7Summaries by Vani AI

OpenAI blocked over 15,000 accounts trying to extract protected model reasoning, linking much of the activity to Moonshot AI.

OpenAI said it disrupted a coordinated campaign to extract protected internal reasoning from its models, taking action against activity linked to more than 15,000 accounts. The effort was first seen in early July 2026 and spiked on July 24 and 25 with about 16,000 requests using one extraction pattern across more than 4,000 users; OpenAI said it disrupted the broader network by July 28. Operators allegedly copied encrypted reasoning from one conversation into another and instructed the model to decrypt and transcribe it. OpenAI said this was adversarial distillation, not a break-in to encryption, internal databases, or stored customer chats, and it attributed a significant portion to individuals associated with Moonshot AI, developer of Kimi, without confirming a single directing actor.

Key facts
  • OpenAI tied the extraction campaign to more than 15,000 user accounts.
  • About 16,000 requests spiked on July 24–25 across more than 4,000 users.
  • The network was fully disrupted by July 28, 2026.
  • One method replayed encrypted reasoning into another chat for transcription.
  • OpenAI linked much of the activity to people associated with Moonshot AI.
Explain this AI news with Vani AIWhat was announced, what is new and why it matters, in a new chat

From the original article

2 min read at gbhackers.com

OpenAI has disrupted a coordinated campaign to extract protected internal reasoning from its AI models on a large scale. The company took action against activity linked to more than 15,000 user accounts. The company described the operation as an adversarial model-distillation effort, in which attackers systematically sought model outputs or reasoning traces that could help […]

GBHackers

By Divya · gbhackers.com

Read the full article at GBHackers

A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.