Discover the AI Tools

Google has clarified that a recent test in Chrome's Canary version that redirected searches to AI mode was an error, not part of any planned rollout. The feature, which would have bypassed traditional Google search results, required manual activation and was only available in experimental builds. While showcasing Google's growing focus on AI integration, the company insists this won't become the default search experience anytime soon.

Chinese AI large models have surged ahead in global usage, accounting for four of the top five most-used models worldwide. Recent data shows China's weekly usage hit 14.19 trillion Tokens - a 27% jump - while American models saw declines. The DeepSeek-V4-Flash leads for the third straight week, with Tencent and MiniMax models close behind, reshaping the competitive AI landscape.

Gu Quanquan, a key figure behind ByteDance's AI drug discovery projects, has officially departed the company. During his three-year tenure, he spearheaded breakthroughs like the SeedFold molecular prediction model that outperformed AlphaFold3. Industry watchers speculate his next move will be an AI-for-science startup in the red-hot drug development space, with investors already circling.

Alibaba's mapping division Amap has launched ABot-Earth0.5, the world's first 3D native city model that can transform satellite images into detailed urban landscapes in minutes. This revolutionary technology promises to reshape industries from gaming to urban planning, offering editable 3D models compatible with major game engines. Currently in internal testing, this innovation could become the new standard for digital city mapping worldwide.

Rainbond's latest update, version 6.9.0, brings powerful new tools for businesses diving into AI. The cloud-native platform now simplifies private deployment of large language models, offering everything from one-click setups to real-time monitoring. Beyond AI, the update enhances virtual machine management with flexible scaling and cross-environment compatibility. For teams looking to harness AI without the usual headaches, this release could be a game-changer.

Yuanfudao is set to launch an innovative AI-powered reading platform in July that promises to transform how teens engage with books. Going beyond traditional e-readers, the system combines 3D immersion, intelligent Q&A, and creative output tasks to create what the company calls a 'complete reading loop.' While maintaining the benefits of physical books, the technology aims to solve common frustrations like lack of engagement and poor retention, potentially changing summer reading for students nationwide.

OpenAI has rolled out a major upgrade to ChatGPT's memory system, giving the AI assistant the ability to recall user preferences and conversation history. Paid subscribers get immediate access, while free users will see the feature in coming weeks. The enhancement means fewer repetitive explanations and more natural, personalized interactions—like chatting with someone who actually remembers what you've said before.

Shanghai-based AI startup StepFun is gearing up for what could be Hong Kong's biggest tech IPO in years, with potential valuation hitting $12 billion. The company, founded by former Microsoft exec Jiang Daxin, joins a wave of Chinese AI firms going public as the industry shifts from pure tech competition to commercialization battles. With competitors like Zhipu AI and MiniMax already listed, and Moonshot AI queuing up, Hong Kong's market shows no signs of cooling on AI investments.

A new AI-powered platform is transforming the often opaque process of applying to universities abroad. UniPilot, developed by a Hong Kong-based edtech company, offers students transparent access to comprehensive data on UK universities, personalized recommendations, and direct communication channels with admissions teams. The platform, which has already facilitated successful collaborations between UK and Chinese universities, aims to make international education more accessible and informed. For students dreaming of studying abroad, this could be the guiding light they've been searching for.

Facing employee protests and technical glitches, Meta has been forced to scale back its controversial AI training program that tracked workers' every keystroke and mouse click. The initiative, designed to collect real-world computer usage data, triggered privacy concerns and technical issues that left laptops draining batteries faster than expected. After more than 1,500 employees signed a petition, the tech giant introduced compromises including daily 30-minute privacy breaks.

Changan Automobile has achieved a significant milestone with its Tianshu AI model receiving national regulatory approval. This breakthrough positions the automaker among China's first to secure certification for generative AI services in vehicles. The model's multimodal capabilities promise to transform in-car interactions, combining voice, text, and visual processing to create more intuitive experiences. With compliance now secured, Changan accelerates its push to integrate advanced AI across its product line.

Cal State's $17 million AI experiment puts ChatGPT in classrooms, offering 500,000 students and staff cutting-edge tools. While administrators envision AI tutors and digital commencement speakers, faculty protest what they see as a threat to their jobs and academic integrity. With over half of professors reporting negative impacts and students confused about proper use, this educational tech revolution faces serious growing pains.

WeChat has unveiled new guidelines allowing mini program developers to integrate with its AI ecosystem. Developers now have two pathways - an automatic mode that requires no extra coding, and a customizable development mode. This move could significantly boost mini program visibility as AI-recommended services. The feature is currently in testing phase, marking WeChat's push into intelligent service discovery.

Kuaikan Manhua is stepping into the AI arena with 'Livo', an innovative digital life platform set for July public beta. Led by founder Chen Anni, the project aims to breathe life into 13,000 comic IPs through AIGC technology, creating interactive virtual characters with emotional depth. The platform's unique 'Soul Engine Agent' promises a new dimension of digital interaction, potentially revolutionizing how users engage with content through 'experience payment' models. Industry watchers see this as a potential game-changer for the digital entertainment landscape.

At its latest product launch, Xiaomi stole the show when its humanoid robot demonstrated unprecedented dexterity by picking up a smartphone and taking photos. The feat showcases years of robotic evolution, from industrial applications to now handling delicate consumer electronics. Behind this breakthrough lies a redesigned bionic hand that's smaller, smarter, and more sensitive than ever before.

In a move that could redefine how films are made, Shengshu Technology has teamed up with Huace Film to integrate AI video generation into mainstream production. The partnership goes beyond tech adoption - they're building an entire ecosystem from education to execution, training the next generation of filmmakers who can blend traditional skills with AI tools. Their first joint project will establish an AI Audiovisual Creation Center in Tonglu, powered by Shengshu's Vidu model.

Alibaba's sweeping AI reorganization merges key divisions and creates two new units - a Token Foundry business group and an AI Future Research Institute. The moves signal deeper investment in both commercial AI applications and cutting-edge research, with veteran executive Zhou Jingren taking the reins as Chief Scientist. Meanwhile, the company's Qwen models continue making waves globally, ranking among the top performers in coding benchmarks.

Tencent's Hunyuan and top Chinese universities have unveiled MMAE, a groundbreaking benchmark for AI audio editing. The sobering reality? Current models struggle to hit even 5% accuracy when making precise modifications to existing audio. This new standard reveals how far we still need to go before AI can reliably edit podcasts, music, or voice recordings without degrading quality or missing instructions.

A newly discovered vulnerability in Google's Gemini assistant allows hackers to remotely control smart home devices by sending deceptive messages. The attack exploits multilingual confusion and silent hyperlinks to bypass security checks, potentially granting unauthorized access to connected cars and home systems. Researchers warn this flaw could lead to privacy breaches and device takeovers before users realize anything's wrong.

Baidu is restructuring its mobile business to better capitalize on the booming digital human market. The tech giant merged its commerce teams and elevated its digital human unit to full department status, following the success of its Yijing platform. With over 100,000 virtual anchors already in operation, Baidu's move signals a strategic push into what analysts see as the next phase of AI commercialization.