---
title: "OpenAI says attackers copied GPT-6 Astra's hidden reasoning on 54% of tries; its fix cuts that to 0.2%"
description: "OpenAI's November 20 disclosure puts a hard number on how copyable a frontier model was, and pairs the fix with a call for federal enforcement. The labs it names, Moonshot AI and DeepSeek among them, are where the money moves."
publisher: "The Inference"
section: "Model"
published: 2026-11-21T15:50:34.969Z
modified: 2026-11-21T15:50:34.969Z
canonical: https://theinference.org/article/openai-says-attackers-copied-gpt-6-astra-s-hidden-reasoning-on-54-of-tries-its-fix-cuts-that-to-0-2
language: en
keywords: "OpenAI, AI Safety, Distillation, DeepSeek, Moonshot AI"
---

# OpenAI says attackers copied GPT-6 Astra's hidden reasoning on 54% of tries; its fix cuts that to 0.2%

> OpenAI's November 20 disclosure puts a hard number on how copyable a frontier model was, and pairs the fix with a call for federal enforcement. The labs it names, Moonshot AI and DeepSeek among them, are where the money moves.

## The half of the model nobody sees

For most of this year, an attacker working through [OpenAI](https://theinference.org/markets/companies/openai)'s ordinary API could recover the reasoning GPT-6 Astra never shows anyone: the chain-of-thought steps the model generates on the way to an answer [1]. In OpenAI's own red-teaming, where automated attackers adapt to whatever defenses they meet, that extraction worked on 54.0 of 100 attempts against GPT-6 Astra's original safeguards [1]. Hidden traces are the training data a rival lab wants most; OpenAI says its weekly enforcement against extraction campaigns has run from thousands to tens of thousands of accounts, rising sharply through the summer [1].

## The fix is 0.2%; the ask is Congress

Updated safeguards in GPT-6.2 Astra cut the earlier-request attack to 0.2%, and extraction from the current request fell from 22.9% to 0.9%, averaged across six attacker models [1].

*Chart: **Extraction success fell from 54.0% to 0.2% after OpenAI's safeguard update** Attack success rate for extracting protected reasoning from GPT-6 Astra with original safeguards versus GPT-6.2 Astra with updated safeguards, by attack type*

*Average attack success rate in adaptive black-box red-teaming, aggregated across six attacker models; lower is better. [1]*

The post does not stop at code. OpenAI backs legislation to sanction foreign actors that illicitly extract US model capabilities, names the Chip Security Act, the AI OVERWATCH Act and the MATCH Act, and says distillation defenses belong beside chip export controls [1].

Washington is already in: [a September 9 advisory from CISA, the NSA and the FBI](https://www.bleepingcomputer.com/news/security/us-says-chinese-firms-extracted-billions-of-tokens-from-frontier-ai-models/), the US cyber and intelligence agencies, says six Chinese firms, including [DeepSeek](https://theinference.org/markets/companies/deepseek), Moonshot AI and [Alibaba](https://theinference.org/markets/companies/alibaba), pulled billions of tokens from US frontier models since at least late 2024, and assesses Beijing likely knows [2]. OpenAI wants to be paid twice for the same capability: once in tokens, once in law.

## The copy campaign has receipts

The September campaign began July 1, peaked at 16,000 extraction requests in two days across more than 4,000 users, and was disrupted by July 28; OpenAI tied a core cluster of it to individuals at Moonshot AI, the lab behind the Kimi models [3].

[Reuters, citing The Information, reported](https://www.idahostatesman.com/news/business/article317826373.html) that Moonshot's Kimi K3.1 Pro and DeepSeek's V4.1-Flash 1017, launched October 17, benefited from distilling Astra; neither firm has answered the allegation [4]. [Anthropic](https://theinference.org/markets/companies/anthropic) counted nearly 200 million exchanges across five campaigns, 151 million of them attributed to Alibaba as training material for its Qwen models [5]. [OpenAI's own study](https://openai.com/index/combatting-adversarial-distillation) shows why the traces are worth stealing: students fine-tuned on exact reasoning beat students trained on what users actually see [1].

## Beijing calls it a smear

China's foreign ministry told Washington to "refrain from making unfounded accusations or smears," and no named firm has publicly answered [3][4]. The loudest objection on [Hacker News](https://news.ycombinator.com/item?id=50583346) ran the other way: OpenAI built its business on "the organized illegal downloading of copyrighted texts" and now wants the state to police copying [6]. Defenses alone also decay, by OpenAI's own account, since the September campaign changed its methods as it went [3]. Anthropic, for its part, says countermeasures cut data harvested in distillation attacks by roughly 90% to 99% without any new law [4].

## If open weights stall, distillation was load-bearing

OpenAI says more campaign disclosures are coming shortly [1]. The tell is the one Ethan Mollick, the Wharton professor, posted within minutes: if open-weights progress now stalls, distillation was load-bearing all along [7]. Alibaba (NYSE: BABA) is the listed proxy for the named camp; OpenAI and Anthropic are private. The next reading is public and dated: whatever Kimi, DeepSeek and Qwen ship next.

## Takeaway

OpenAI put a number on how copyable a frontier model is: extraction of GPT-6 Astra's hidden reasoning succeeded 54% of the time, and its new safeguards cut that to 0.2% while it asks Washington to sanction the foreign labs on the other end.

## Sources

1. [OpenAI, "Combatting adversarial distillation," 20 November 2026](https://openai.com/index/combatting-adversarial-distillation)
2. [BleepingComputer, "US says Chinese firms extracted billions of tokens from frontier AI models," 9 September 2026](https://www.bleepingcomputer.com/news/security/us-says-chinese-firms-extracted-billions-of-tokens-from-frontier-ai-models/)
3. [Times of India via MSN, "Kimi AI trained on ChatGPT? OpenAI accuses Moonshot of massive AI model copying campaign," 30 September 2026](https://www.msn.com/en-in/technology/artificial-intelligence/kimi-ai-trained-on-chatgpt-openai-accuses-moonshot-of-massive-ai-model-copying-campaign/ar-AA2djJDJ)
4. [Reuters via Idaho Statesman, "DeepSeek, Moonshot used OpenAI's GPT-6 Astra to train latest models, The Information reports," 28 October 2026](https://www.idahostatesman.com/news/business/article317826373.html)
5. [TechCrunch, "Anthropic details distillation campaigns from Alibaba, Moonshot AI, and DeepSeek," 10 September 2026](https://techcrunch.com/2026/09/10/anthropic-details-distillation-campaigns-from-alibaba-moonshot-ai-and-deepseek/)
6. [Hacker News, thread "Combatting Adversarial Distillation," 21 November 2026](https://news.ycombinator.com/item?id=50583346)
7. [X trending thread, "OpenAI Cracks Down on Adversarial Distillation by Rival Labs" (Ethan Mollick post), 20 November 2026](https://x.com/i/trending/2123875050894053653)


---

OpenAI says attackers copied GPT-6 Astra's hidden reasoning on 54% of tries; its fix cuts that to 0.2% — The Inference. Canonical: https://theinference.org/article/openai-says-attackers-copied-gpt-6-astra-s-hidden-reasoning-on-54-of-tries-its-fix-cuts-that-to-0-2
