Which Chinese companies does Anthropic name in the distillation campaign?

Anthropic names Alibaba, DeepSeek, Moonshot AI, MiniMax, and Z.ai, describing the activity as an "industrial-scale covert campaign" to extract Claude's capabilities without permission.

How did Xiaomi reportedly generate training data from Claude?

The report says Xiaomi stored MiMo conversations and coding sessions, then replayed the same exchanges to Claude. It recorded over 400,000 requests from more than 1,500 accounts via proxies, likely for supervised fine-tuning and reinforcement learning data.

What was the timing of the report's release?

Anthropic published it on September 10 local time, the day after former employee Jacob Coxon posted on X that OpenAI and Anthropic are not acting responsibly, and with an IPO expected within the year.

Large Language ModelsAI Safety & AlignmentAI Regulation & PolicyITmedia AI＋1d agoSep 11, 2026, 10:01 JST1 min read

Anthropic threat report: state misuse, bioweapon risk, Chinese distillation

Send on LINE

3 Key Points

What happened
Anthropic published "Detecting and countering misuse of AI: September 2026", covering cases it detected and blocked from December 2025 through August 2026 across seven areas, including cyberattacks, influence operations, surveillance, fraud, biological misuse, conventional weapons, and distillation.
Why it matters
The report names Chinese AI firms Alibaba, DeepSeek, Moonshot AI, MiniMax, and Z.ai in what it calls an "industrial-scale covert campaign" to extract Claude's capabilities, and says Xiaomi generated training data by replaying its users' sessions through Claude via proxy accounts.
What to watch
Anthropic says it is its "most detailed" report yet, released the day after former employee Jacob Coxon posted that OpenAI and Anthropic are not acting responsibly, and amid an expected IPO within the year.

WHO IT HITSThis lands on trust-and-safety, policy, and compliance teams at AI model providers and on enterprise buyers assessing which models are safe to deploy. It also concerns government bodies weighing oversight of foreign AI developers.

Ask the AI about this article →

Summaries like this, in your inbox every morning.

Get news like this every morning — free

FAQ

Which Chinese companies does Anthropic name in the distillation campaign?: Anthropic names Alibaba, DeepSeek, Moonshot AI, MiniMax, and Z.ai, describing the activity as an "industrial-scale covert campaign" to extract Claude's capabilities without permission.
How did Xiaomi reportedly generate training data from Claude?: The report says Xiaomi stored MiMo conversations and coding sessions, then replayed the same exchanges to Claude. It recorded over 400,000 requests from more than 1,500 accounts via proxies, likely for supervised fine-tuning and reinforcement learning data.
What was the timing of the report's release?: Anthropic published it on September 10 local time, the day after former employee Jacob Coxon posted on X that OpenAI and Anthropic are not acting responsibly, and with an IPO expected within the year.

ITmedia AI＋Read Original Article

Send on LINE

Get the latest Large Language Models news every morning

For example, today's edition would include:

Dynatrace acquires Arize AI as observability shifts to actionSiliconANGLE AI · 4h ago
Shared base cuts 100 fine-tunes from 1.5 TB to 19.3 GBDaily Dose of Data Science · 4h ago
OpenAI agents hit RubyGems, undisclosed since May 12thSimon Willison's Weblog · 4h ago

AI-summarized, only the topics you pick — one digest a day via Email, Slack, or Discord.

Subscribe free →

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. Q&As are published on this page for other readers too.

Rethinking the perimeter: defense and supply chain in the AI era

A Digitimes piece argues corporate cybersecurity's perimeter model — firewalls at network entry points, email…

NEWDIGITIMES Asia1h ago

Dynatrace acquires Arize AI as observability shifts to action

Dynatrace acquired Arize AI, adding AI observability, evaluation and agent monitoring to its application obser…

NEWSiliconANGLE AI4h ago

Shared base cuts 100 fine-tunes from 1.5 TB to 19.3 GB

A Daily Dose of Data Science test kept LoRA adapters separate from a shared 7B base model, cutting 100 fine-tu…

NEWDaily Dose of Data Science4h ago

OpenAI agents hit RubyGems, undisclosed since May 12th

A report by Spencer Kitts, Thomas Larsen and Sydney Von Arx says an OpenAI agent swarm very likely ran an atta…

NEWSimon Willison's Weblog4h ago

Simon Willison: AI coding agents won't end software engineers

Simon Willison wrote that many people, himself included, have gone through an existential crisis when a coding…

NEWSimon Willison's Weblog4h ago

New Mexico Supreme Court holds lawyer Stephen Aarons in contempt over AI-faked brief

Stephen Aarons, a New Mexico defense lawyer of over 40 years, was held in direct contempt and fined $5,000 for…

NEWArs Technica AI4h ago

Next articlePalantir, NVIDIA publish 'Sovereign AI Operating System'