🔍 Read the full analysis: How China’s New AI Models May Relate To GPT-6 Astra, According To A Report on ThorstenMeyerAI.com
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
TL;DR
The Information reported on Oct. 28 that DeepSeek and Moonshot AI relied heavily on OpenAI’s GPT-6 Astra outputs to train recently released models. Reuters reported that none of the three companies responded to its requests for comment; public details about the alleged data, methods and scale remain limited.
The Information reported Oct. 28 that Chinese AI labs DeepSeek and Moonshot AI relied heavily on outputs from OpenAI’s GPT-6 Astra to train recent models, a claim that could raise questions about model competition and OpenAI’s terms of service. Reuters relayed the report, but none of the three companies had confirmed or denied it in responses to Reuters as of its publication.
The models named are Moonshot’s Kimi K3.1 Pro, released Oct. 23, and DeepSeek-V4.1-Flash-1017, released Oct. 17, according to Reuters’ summary of The Information. The original report, by Jing Yang, Juro Osawa and Qianer Liu, said Astra had become a growing source for Chinese AI labs seeking to extract capabilities from the model. The article’s public opening is available, but its full reporting is paywalled.
The account rests on unnamed sources described as having direct knowledge, according to the reports. Publicly available coverage does not specify how many Astra outputs were used, how the labs obtained them, or what parts of the models’ training were affected. Moonshot’s Kimi launch post attributed its gains to substantially more post-training compute and large-scale reinforcement learning. DeepSeek’s update note did not describe its training data. Those statements do not independently resolve the allegation.
Reuters said OpenAI, DeepSeek and Moonshot did not immediately respond to its requests for comment. A separate summary by South Korean broadcaster SBS attributed a statement to an OpenAI spokesperson saying the company takes the issue seriously and continues to work on it. That detail was not independently checked against The Information’s full report in the source material. The central claim therefore remains a reported allegation, not a public finding established by disclosed records.
Infographic 01 · Distillation in plain words
How distillation works, and where it crosses a line.
A big model teaches a small one by example. The technique is ordinary; the dispute is about whose model is the teacher and how the answers were obtained.
Send the teacher prompts
A strong “teacher” model gets a huge number of carefully chosen tasks.
Keep answers and reasoning
The step-by-step working is worth far more than the final answer alone.
Teach a student to imitate
A smaller or cheaper “student” model learns to reproduce the teacher’s moves.
Normal practice
The U.S. advisory calls it “a legitimate and useful technique”.
- A lab distills its own large model
- Goal: smaller, faster, cheaper versions
- Open models used within their licence
Where the fight starts
What OpenAI, Anthropic and U.S. agencies allege.
- The teacher is a rival’s model
- Industrial scale: millions of exchanges
- Fake, resold or stolen accounts and keys
- Breaks terms that ban building competing models
Why hidden reasoning is the prize
Anthropic’s own test: a model trained on Claude Opus 4.8 transcripts, scored on SWE-bench Verified (coding).
That is why providers now hide or summarize reasoning, and why extraction attempts target it.
Distillation is a student copying a top tutor’s worked solutions. Legal in a classroom. Contested when the tutor never agreed.Thorsten Meyer · my reading
Sources: NSA/CISA/FBI advisory AA26-251A, 8 Sep 2026; Anthropic, 12 Oct 2026 (vendor’s own experiment). Analogy and framing: my reading.
Powered by Thorsten Meyer AI
Why Training Provenance Matters
If the report is accurate, it would put a familiar AI training technique at the center of a dispute over whose outputs can be used and under what conditions. Model distillation involves training one model to imitate responses from another. Labs commonly use it with their own systems; using a competitor’s model at large scale could raise contractual questions if it violates the provider’s terms.
The commercial stakes are clear in the price and performance figures cited by the source article. It lists Astra at $10 per million input tokens and $50 per million output tokens, compared with Kimi K3.1 Pro at $2 and $10, and DeepSeek’s model at $0.30 and $1.20 during peak hours. On the same independent index, the three models scored 57, 46 and 32, respectively. The figures suggest lower prices for the named models, but also lower scores on that index; they do not establish why the models differ or whether distillation explains any of the results.
For businesses choosing AI systems, the dispute is a reminder to ask providers about training-data provenance and access controls, while testing models on their own workloads and keeping alternatives available. For users, the report does not establish that anything has changed in the named apps. The source article advises caution with cheap, unofficial API resellers, but provides no evidence that a particular reseller was involved in the alleged activity.
Products Worth Considering
Apple iPad Air (2025) 13‑inch Wi‑Fi 256 GB – Space Gray (M3 chip)
DJI Osmo Pocket 3 Creator Combo Vlogging Camera with Mic 2, 1 CMOS, 4K/120fps, 3-Axis Stabilization, 128GB Card
DJI Osmo Pocket 3 Vlogging Camera with 1-Inch CMOS Sensor and 4K Video
As an affiliate, we earn on qualifying purchases.
Distillation and Earlier Allegations
Distillation is a standard research technique: a smaller “student” model is trained using answers from a larger “teacher” model. The U.S. government advisory referenced in the source material describes distillation as a legitimate and useful technique in AI research. The contentious issue is not the technique alone, but whether a lab used a rival’s model in a way that breached its rules or involved deceptive access. No public evidence cited in the report establishes the method allegedly used by DeepSeek or Moonshot.
The report comes amid earlier accusations involving attempts to extract model capabilities. The source article says OpenAI tied a “core cluster” of a July campaign targeting hidden reasoning to individuals associated with Moonshot AI, and that U.S. security agencies named both DeepSeek and Moonshot in an advisory. It also says China’s commerce ministry called accusations baseless. These are separate claims and responses; they do not independently verify the new report about Astra or establish responsibility for the alleged training.
The source material also cites a February 2026 Anthropic account of large-scale exchanges with Claude involving DeepSeek, Moonshot and MiniMax. Such prior allegations provide background to current concerns about access and model imitation, but the details supplied here are not enough to establish a direct link between those events and the Astra report.
What the Public Record Does Not Show
The available reporting does not disclose the number or content of Astra outputs allegedly used, the accounts involved, or the training process. It also does not show how much any such data contributed to either model’s capabilities. The original article is largely behind a paywall, and Reuters’ account relies on the reported findings rather than published supporting records.
There is also no public response from DeepSeek or Moonshot in the material provided, and no detailed OpenAI statement directly addressing the report. Without company responses, documented access records or other independently verifiable evidence, it is not possible to determine whether the alleged activity occurred as described, whether it breached a contract, or whether any legal violation took place.
Responses and Evidence to Watch
The next developments to watch are substantive responses from OpenAI, DeepSeek and Moonshot, along with any additional reporting that documents the scale and method of the alleged training. Company statements, technical evidence or disclosed contractual action could clarify whether the dispute concerns ordinary access to model outputs, alleged use at industrial scale, or the way access was obtained.
Until more evidence is public, the report should be treated as an unconfirmed allegation attributed to unnamed sources. The models remain available, according to the source material, and no immediate change to their apps or services is reported. Business users can review provider terms and test alternatives while the facts develop; that is prudent procurement practice, not confirmation of wrongdoing.
Key Questions
What does the report say DeepSeek and Moonshot did?
The Information reported that both labs relied heavily on GPT-6 Astra outputs to train recent models. Reuters relayed the claim; the companies have not confirmed it in the source material.
Which models are named?
The report names Moonshot’s Kimi K3.1 Pro, released Oct. 23, and DeepSeek-V4.1-Flash-1017, released Oct. 17.
Is distillation illegal?
Not inherently. Distillation is a common training technique. Whether a particular use violates a provider’s terms or law depends on the facts, including how access was obtained and what rules applied. The reporting does not establish a legal violation.
Have the companies confirmed the allegation?
No. Reuters reported that OpenAI, DeepSeek and Moonshot did not immediately respond to its requests for comment. The source material provides no detailed public confirmation or denial from the companies.
Does the report mean users need to stop using these models?
The report does not document an immediate change to the models’ apps or establish that users’ accounts or data are affected. Businesses may wish to review provider terms and test alternatives, while users should distinguish the reported allegation from confirmed findings.
Source: ThorstenMeyerAI.com
