Anthropic identifies extensive AI model 'distillation' by China's DeepSeek, MiniMax, Moonshot

Entertainment National Sports Markets Business & Economy Lifestyle World Astrology & Religion Technology Education & Jobs Auto Politics Viral Opinions

Loading more articles...

Anthropic Flags Chinese AI Firms for 'Industrial-Scale Distillation' of Claude Models

No 4. Company: Anthropic | Valuation: $183 billion | Country: US | Anthropic, which set shop in India in October 2025 is an AI research company, which focuses on building safe and interpretable AI systems. Valued at $183 billion, it is one of the fastest-rising players in the artificial intelligence sector. The firm has gained attention for its Claude AI model, seen as a key rival to OpenAI’s offerings.

C

CNBC TV18•24-02-2026, 09:17

Anthropic Flags Chinese AI Firms for 'Industrial-Scale Distillation' of Claude Models

•US AI company Anthropic alleges DeepSeek, MiniMax, and Moonshot AI conducted large-scale 'distillation' campaigns.
•Over 16 million interactions with Claude were generated using roughly 24,000 fraudulent accounts, breaching terms of service.
•Distillation, a common AI training method, was allegedly misused to replicate advanced capabilities at lower cost and time.
•DeepSeek focused on reasoning tasks, Moonshot AI on agentic reasoning and coding, and MiniMax on agentic coding and orchestration tools.
•Anthropic warns of growing intensity and sophistication of such attacks, calling for coordinated industry and policymaker action.

Read Full Article on Cnbc in English

✦

More like this

✦

More like this

Chinese AI Companies Allegedly Exploit Anthropic’s Claude Model for Advanced Capabilities

Anthropic Accuses Chinese AI Firms of Illicit Claude Model Exploitation

OpenAI + Ginkgo Lab Loop: OpenAI connected GPT-5 to an autonomous laboratory setup that autonomously proposes experiments, executes them, and learns from results without human intervention. This closed-loop system accelerates scientific research by reducing reliance on manual experimental planning.

OpenAI Accuses China's DeepSeek of Illegally Distilling US AI Models

PM Narendra Modi (centre) with Sundar Pichai (left), Sam Altman and Dario Amodei and others at the AI Summit. In the picture, Altman and Amodei (both on the right) can be seen not holding hands.

AI Rivalry Heats Up: Altman, Amodei Avoid Hand-Holding at AI Summit

Claude AI

Anthropic Upgrades Claude Sonnet to 4.6: Stronger Coding, Agents, and Long-Context Reasoning

Anthropic researchers observed that the model sometimes lost control during training. (Representational)

Anthropic Warns: New AI Model Claude Opus Can Misbehave, Act Without Human Permission

Representational image (Image: News18)

Agnikul to launch AI data center prototype in orbit by year-end, aims for 2027 commercialization