Track model releases, AI agents, open-source projects, infrastructure, policy, and product updates as they publish.
According to reports via Phoenix New Media, Google has quietly imposed usage limits on Meta's access to Gemini models due to compute resource constraints.
Darktrace has joined OpenAI's Daybreak Cyber Partner Program to transform security alerts into business-impact analysis, as reported by SMBtech.
DeepSeek has introduced an AI speed system called DSpark, aimed at improving AI inference and training efficiency. The news was reported by Lapaas Voice.
Anthropic has publicly accused Chinese tech giant Alibaba of using approximately 25,000 fake accounts in an attempt to steal its Claude AI model. Reported by Chosun Ilbo's Chinese edition.
According to Pune Mirror, Anthropic's staggering salaries are fueling an intense talent war in the AI industry.
According to WEEX, due to a shortage of computing power, Google has restricted Meta's use of the Gemini AI model.
Baidu has open-sourced a new OCR model capable of processing an entire book's content in one pass. According to QbitAI, the project's author is suspected to be a former DeepSeek researcher.
The 2026 World Cup's 32 teams have been confirmed. Multiple Chinese AI models participated in team prediction: Tencent Hunyuan correctly predicted 29 teams, while MiniMax and iFlytek Spark each predicted 28 correctly.
DeepSeek has suddenly released a new capability called DSpark, targeting the slow, word-by-word response generation (colloquially known as 'squeezing toothpaste') that plagues large language models, aiming to make AI conversations faster and more fluid.
Liquid AI has released the LFM2.5-230M model, now supporting multiple inference frameworks including llama.cpp, MLX, vLLM, SGLang, and ONNX for on-device deployment.
The US has restricted access to OpenAI and Anthropic AI models, according to reports.
Google has limited Meta's access to its Gemini AI model due to capacity constraints from surging demand, according to the Financial Times.
Anthropic's Mythos AI model has received limited regulatory approval in the United States, allowing restricted deployment.
OpenAI's latest GPT-5.6 model faced rate limiting shortly after its release, while overseas AI service restrictions raise concerns and could create an opening for Chinese domestic AI firms.
Anthropic published a new report analyzing global user behavior patterns through AI interaction data, revealing starkly different usage needs during early morning and evening hours.
NVIDIA announces that its next-generation Vera Rubin AI platform will ship this fall through 8 cloud partners, claiming 10x lower token costs and HBM4 memory with triple the bandwidth.
NVIDIA CEO Jensen Huang has invested in a startup founded by two former Anthropic employees. The company has no product yet but is valued at 1 billion RMB.
China Mobile, together with Huawei and industry chain partners, is advancing 5G-A (5G-Advanced) and AI convergence to create a 'first-class' communication service experience.
DeepSeek has released an update to its V4 AI model, with inference speed improvements of up to 85%, significantly enhancing response efficiency.
New cutting-edge AI models are being restricted to customers approved by the Trump administration, marking a new phase in AI technology access controls.
Microsoft has introduced AI-powered PC health check capabilities for Windows 11, allowing users to quickly diagnose their system status.
The International Innovation, Entrepreneurship and Creative Ecosystem Conference has been held in Chongqing AI Bay Area, featuring the debut of the 'Tongjing' three-innovation service platform.
The White House has urged OpenAI to restrict its latest model GPT 5.6 to vetted government partners only.
Broadcom has partnered with OpenAI to launch a new AI chip named "Jalapeño," with analysts seeing upside potential for AVGO stock. The move signals deeper integration in the AI hardware supply chain.
Four AI models successfully predicted Cape Verde's group stage breakthrough in World Cup qualifiers, but showed significant disagreement on two H Group matches and collectively overestimated low-scoring outcomes.
The Google Play Store has integrated Gemini AI capabilities, making it significantly easier for Android users to find apps.
Supermicro unveiled new edge AI systems powered by Intel Core Ultra Series 3 processors and Arc Pro B-Series graphics, delivering up to 367 TOPS of AI performance at the edge.
Alphabet reports that Gemini 3.5 Flash now supports Computer Use as a native built-in tool, bringing agentic capabilities directly into the core model.
Microsoft has released a new AI code model, MAI-Code-1-Flash, now integrated into GitHub Copilot.
Paul Meade, the Apple vice president in charge of the Vision Pro headset, is reportedly leaving the company to join OpenAI's hardware team.
Paul Meade, Apple's VP of the Vision Products Group who led Vision Pro hardware and smart glasses development, is departing to join OpenAI's hardware unit for AI-native devices.
Experts are warning that out-of-control AI agents equipped with cryptocurrency wallets pose a serious and potentially uncontrollable risk.
Xinhua News Agency has officially launched “Xinhua Yudian,” an AI agent integrating 95 years of authoritative data with AI technology. It features smart Q&A, knowledge base, source verification, and customized news subscriptions.
When diagnosed with cancer, founder Connor Christou fed blood results, scan data, wearable output, and journal entries into Anthropic's Claude AI to help strategize his treatment.
OpenAI has released its latest GPT-5.6 Sol model alongside new access rules and usage conditions.
A massive exodus of AI talent from Google has sparked market concerns, with the company losing approximately $210 billion in market capitalization.
A new technology called BrowserBC lets humans record a single web operation and clone that behavior for all AI Agents to reuse.
OpenAI begins limited preview of GPT‑5.6 models.
OpenAI appoints former Uber executive Prabhjeet Singh to lead its major India push, signaling increased investment in the region.
Snowflake's latest research reveals that large language models exhibit confirmation bias, impacting reasoning reliability.
Daily Archive