Discover the Latest AI News📰

Major tangyuan brands are reminding customers that March 3rd marks the final day to redeem Qwen's free meal cards, originally issued during Spring Festival. The 25-yuan digital coupons can be used for both packaged tangyuan and restaurant orders, dropping prices to just 0.01 yuan. Meanwhile, Qwen's expanded 'Super Inviting Card' offers broader spending options until April 30th. The promotion comes as Qwen reports staggering growth, with February seeing 203 million active users globally.

DeepSeek's V4 Lite model, initially overshadowed by its upcoming full version, has quietly evolved into a formidable competitor in the AI space. With just 200 billion parameters - modest by today's standards - it now rivals top international models like Claude 3.5 Sonnet in performance. Recent upgrades have particularly impressed developers with its enhanced logic, programming capabilities, and aesthetic judgment.

Telegram's latest Bot API 9.5 update brings game-changing streaming capabilities to all chatbots, letting them display responses in real time just like ChatGPT. OpenClaw has already adapted this feature, making it the first framework to fully support Telegram's new native streaming. This eliminates clunky workarounds developers previously used, promising smoother conversations across private chats, groups and topics.

Kuaishou has removed over 4,000 videos featuring inappropriate AI modifications of classic films and animations. The crackdown focuses on protecting minors from disturbing content and preserving the integrity of cultural treasures like 'Journey to the West.' The platform vows to strengthen AI detection while calling on users to help report violations.

Ant Digital's Tianji Lab has revolutionized embodied intelligence training with their AoE framework. By transforming ordinary smartphones into powerful data collectors using simple neck mounts, they've slashed costs from thousands to just $20. The system captures natural first-person interactions with surprising accuracy, then automatically processes footage into usable training data. Early tests show dramatic improvements - robot success rates jumped from 45% to 95% when supplemented with AoE-collected data.

Zopia has launched as the world's first end-to-end AI video director, transforming how short films are made. With just a text prompt, this innovative system handles everything from scripting to final edits using advanced multi-agent coordination. Optimized for leading video models Kling3 and Vidu Q3, Zopia solves longstanding challenges like character consistency and scene continuity. Early adopters report professional-quality outputs that could disrupt content creation across platforms.

Alibaba's DAMO Academy has partnered with Pakistani hospitals and tech firm Sky47 to introduce AI-powered multi-cancer screening technology. The system, already serving 20 million people globally, helps detect subtle lesions missed by human eyes. This collaboration represents China's growing medical AI exports along Belt and Road nations.

Meituan's tech team Guangnian Zhiwai has addressed allegations of code copying in its Tabbit AI browser. After developers flagged similarities with open-source project Read-Frog, the company removed the disputed translation features and released the code publicly. This incident highlights growing tensions between rapid AI development and open-source compliance.

WeChat cracked down hard last month on accounts misusing AI tools to distort classic films and animations. Nearly 4,000 videos were removed for vulgar adaptations of literary classics like 'Romance of the Three Kingdoms' and creepy makeovers of children's cartoon characters. The platform vows to maintain strict oversight as part of its long-term content moderation strategy.

Chinese AI firm MiniMax has reported explosive growth in its first annual report since going public, with revenue jumping 159% to $79 million. While the company posted massive paper losses due to valuation changes, its core AI platform shows promising traction globally - especially with its cost-effective M2.5 model that's gaining ground against competitors. However, copyright lawsuits from Hollywood studios loom as potential risks for this rising AI star.

As AI chatbots like ChatGPT become the new search engines, brands face a critical challenge: staying visible when users ask for recommendations. Traditional SEO is no longer enough - Generative Engine Optimization (GEO) is the new battleground. This article explores why domestic AI platforms matter more than you think, examines the current GEO tool landscape, and highlights a comprehensive solution for tracking brand performance across Chinese AI platforms.

Meta is quietly testing a new AI shopping feature that leverages its vast trove of social data to deliver hyper-personalized product recommendations. Unlike traditional search, the tool acts like a personal shopper that understands your preferences based on location, browsing history, and even inferred gender. While still in early stages, this move signals Meta's ambitions to compete with Google and OpenAI in the AI commerce space.

Meta is quietly testing a new AI shopping feature that could change how we discover products online. The tool offers personalized recommendations in a visual carousel format, pulling from your location, preferences, and browsing history. While currently limited to some U.S. users, this move signals Meta's ambition to compete with tech giants in the AI commerce space.

Chinese tech company Rokid has upgraded its AI glasses to support four major language models simultaneously - Google Gemini, ChatGPT, DeepSeek and Alibaba's Qwen. The lightweight smart glasses weigh less than regular sunglasses while packing serious AI capabilities including real-time translation, object recognition and creative assistance. This move positions Rokid as a strong competitor against Meta's Ray-Ban smart glasses.

Anthropic's AI assistant Claude faced a perfect storm last week - servers buckling under unprecedented demand while caught in a political crossfire. The platform saw user numbers skyrocket just as former President Trump banned federal use, calling the company 'too woke.' Meanwhile, leaked documents reveal the military continues relying on Claude for sensitive operations, creating an ethical dilemma that pits national security against AI principles.

At MWC 2026, iFLYTEK unveiled groundbreaking AI glasses weighing just 40 grams that revolutionize cross-language communication. These smart glasses display real-time translated subtitles while playing audio translations, creating a seamless 'simultaneous interpreter' experience. Their innovative lip-reading technology cuts through noisy environments, improving accuracy by over 50%. Available for pre-order starting March 4th.

Alibaba's Qwen team has unveiled a series of lightweight AI models that pack impressive capabilities into small packages. These new models, ranging from 0.8B to 9B parameters, offer multimodal processing while being optimized for edge devices like smartphones and IoT gadgets. The smallest models deliver lightning-fast performance, while the larger ones rival much bigger systems in capability - all while consuming fewer resources. Available now on popular platforms, these models could revolutionize how we deploy AI in everyday devices.

At MWC 2026, iFLYTEK unveiled groundbreaking AI glasses that promise seamless multilingual conversations. Weighing just 40 grams, these smart glasses combine lip-reading tech with advanced translation to cut through noisy environments. The device projects real-time subtitles while maintaining natural eye contact - potentially revolutionizing international business and travel communication.

A recent accidental code leak suggests OpenAI's upcoming GPT-5.4 model will feature unprecedented memory capabilities, including a 2 million token context window and true stateful AI functionality. The leaked details indicate this could transform AI from temporary assistants to persistent digital colleagues, remembering workflows across sessions. While OpenAI quickly attempted to cover the leak by rebranding it as GPT-5.3-codex, tech analysts believe this reveals their strategic move against competitors like Claude and Gemini.

ChatGPT saw a staggering 295% spike in U.S. uninstalls after OpenAI's defense deal became public, while rival Claude gained traction by refusing similar partnerships. The backlash highlights growing consumer concerns about AI ethics in military applications.