China Opens Probe Into DeepSeek and Moonshot AI Over Suspected Leak of Classified Data

โดย Jiji Press·CNUS·Read original
Summary · why it matters

China's Cyberspace Administration has launched an investigation into Chinese startups DeepSeek and Moonshot AI over suspicions that state secrets were leaked to U.S. AI companies. The U.S. tech outlet The Information reported on the 22nd that Chinese authorities began the probe amid concerns over "distillation," in which the outputs of high-performance American AI models are used to train Chinese AI systems. According to a report published by U.S. AI startup Anthropic, DeepSeek and Moonshot sent users' questions to Anthropic's cutting-edge Claude model without authorization and used its answers to train their own models. The data involved also included internal information from the Chinese People's Liberation Army, public security authorities, and state-owned enterprises. The Trump administration and U.S. national security and investigative authorities have sharply condemned distillation. Ahead of the U.S.-China summit on the 24th, the issue of data security has come into sharp relief.

Theme Impact 2

Off-coverage companies 1

Moonshot AI (北京月之暗面科技有限公司)Private▼ Negative
Regulationrelevance

China's Cyberspace Administration launched a probe into Moonshot AI over suspected leak of classified data via distillation.

Related news

3

Anthropic Unveils Claude Opus 5.5 as OpenAI Releases Terra and Luna Models

Anthropic unveiled its new Claude Opus 5.5 on Tuesday, a model the developer says offers performance approaching its frontier Fable 5.1 at a lower cost for enterprise customers. OpenAI also just released its new GPT-6 Sol and Luna models, which are its lower-end models outside of Solar and Astra that it previously had. The twin launches underscore a shift across the AI industry toward cheaper offerings as companies rein in spending on tokens and shop around for models that fit their needs and budgets. Anthropic included notes on security in its release announcement, while OpenAI has said it is working with third-party external partners to audit its AI models. The push comes amid debate over AI safety and a lawsuit alleging a group of large companies colluded to slow AI development and degrade performance for paying consumers.
Yahoo Finance·1hRead more →
11impact 4

Alibaba Unveils Zhenwu V900 AI Chip, Plans 10-Trillion-Parameter Model

Alibaba Group shares climbed about 3% Tuesday after the Chinese technology company introduced a new in-house processor and outlined plans for a much larger AI model. The company said its next-generation model could reach 5 trillion to 10 trillion parameters, well above its current flagship Qwen model. Alibaba also unveiled the Zhenwu V900 through its T-Head semiconductor unit, a processor designed to deliver three times the performance of its predecessor and scheduled to enter mass production in the first quarter of 2027. The hardware push is tied to a broader expansion of Alibaba's computing infrastructure, with Alibaba Cloud planning to take its global data-center capacity beyond 20 gigawatts by 2032. The moves come as Chinese technology companies seek greater control over AI computing amid U.S. restrictions on advanced processors, with Alibaba's plans pointing to a strategy spanning chips, models and cloud infrastructure rather than relying on a single layer of the AI stack.
GuruFocus·3hRead more →

Meta Rallies as Muse AI Chatbot Tops App Store Downloads

Meta Platforms stock is trying to hold onto gains from a rally sparked by the release of its new Muse AI chatbot, which yesterday sent the market buzzing up like 11% and became the number one downloaded app in the App Store. On the Today's Market Hang panel, Interactive Brokers chief strategist Steve Sosnick said the stock was actually the most sold on the Interactive Brokers platform as customers took profits and sold into the rally. Gerber Kawasaki Wealth & Investment Management CEO Ross Gerber said he does not use Instagram or any Meta product, citing trust concerns and asking why anyone would give Muse access to their phone and personal information when Facebook cannot be trusted. Sosnick echoed that concern, saying he does not need more Facebook, more Meta, or more Zuckerberg in his life. Gerber argued trust is a main factor people are discounting for AI agents, and that Apple holds the pole position if it can build something better than Apple Intelligence. Defiance ETFs CIO Sylvia Jablonski said Meta has an enormous structural advantage given its existing user base and digital footprint, making Muse a service within behavior users already have, though she agreed trust remains the issue.
Yahoo Finance·4hRead more →