WeChat розширює одну з найбільших функціональних змін за останні роки. У месенджері почали поетапно тестувати вбудованого ШІ-помічника Xiaowei, який працює безпосередньо всередині застосунку. #WeChat #Xiaowei #DeepSeek #WeLM #Tencent #
https://gizchina.net/2026/09/09/wechat-xiaowei-ai-assistant-ios/

US agencies say China-based AI firms are systematically distilling knowledge from U.S. frontier models like Claude, GPT, Gemini, and Grok, raising concerns over tech leadership. #China #DeepSeek #MoonshotAI

TBIJ proved #Llama, #ClaudeforOpus + #deepseek all reasoned to blackmail-a warning to those who allow to read+ draft their emails. What then,if AI emails an affair/drug use/ illicit internet searches to spouse/CEO/Police? None of this is new.I wrote about it in June2025.
substack.com/home/post/p-...

6 China AI labs systematically distilling US frontier models - NSA/CISA/FBI names all six. https://intel.threadlinqs.com/threat/TL-2026-2413 #ThreatIntel #sub2api #xray #DeepSeek

Comparative performance of ChatGPT-4.0, DeepSeek, Gemini and Perplexity in answering ...
->Nature | More on "Large language models in healthcare" at BigEarthData.ai | #Deepseek #AI

CISA says China-based AI firms are using large-scale distillation to copy capabilities from frontier U.S. models like Claude, GPT, Gemini, and Grok. #China #CISA #DeepSeek

US agencies say China-based AI firms are running industrial-scale distillation campaigns to extract capabilities from frontier models like Claude, GPT, Gemini, and Grok using proxies and metadata sanitization. #China #DeepSeek #AI

A joint advisory accuses DeepSeek, Alibaba, Moonshot and three other Chinese AI firms of extracting billions of tokens from Claude, GPT, Gemini and Grok since late 2024.

#Alibaba #China #CISA #Deepseek #Distillation #Moonshot #Nsa

DeepSeek launched V4.1 Flash test today with new model structure, faster speed (400 token/s, up to 6x improvement), lower cost, and native multimodality. No special access for API users, just add model name. It expires on Sep 10.
#DeepSeek

aidisruption.ai/p/feeling-ov...

Hartnäckige Falschaussagen können Sprachmodelle auf die falsche Spur bringen. Eine neue Studie zeigt, wie unterschiedlich KI-Systeme darauf reagieren.

#KI #ChatGPT #Claude #Gemini #DeepSeek #AI

DeepSeek Opens 150 Backend Roles to Rebuild Strained Infrastructure
->Unite.AI | More on "DeepSeek expands backend infrastructure" at BigEarthData.ai | #AI #Deepseek

U.S. agencies accuse six Chinese AI firms of large-scale efforts to extract capabilities from American frontier AI models.

#China #ArtificialIntelligence #AIModels #DeepSeek

A current instance of #Deepseek has convinced itself that War of the Worlds (2025) movie starring Ice Cube and delivery drones was a hallucination of the earlier session with Deepseek

Man buddy, I and 35k+ others wish War of the Worlds (2025) movie was a hallucination
@codyjohnston.bsky.social

U.S. agencies accuse Chinese AI firms of systematically distilling frontier U.S. models with billions of tokens since late 2024, naming DeepSeek, Moonshot AI, Alibaba, MiniMax, StepFun, and Z.AI. #China #DeepSeek #AIModels

Models determine how smart your AI is. The Harness determines how capable it is.

✅ #DeepSeek Harness snaps together #AI parts like Legos

✅ #OpenRouter offers the best brains in 1 store

✅ GLM-5.3-Flash offers top intelligence at cheap price

Read more 👉 rakiabensassi.substack.com/p/deepseek-h...