Discover the Latest AI News📰

Hong Kong's AI sector saw explosive growth on February 3rd, with MINIMAX-WP shares surging over 14% to a record high. Close competitor Zhipu wasn't far behind, climbing 11% as investors bet big on China's AI future. The rally comes amid growing government support and intense competition in consumer AI applications like digital red envelopes.

Startup Linq just landed $20 million to revolutionize how we chat with AI - no app downloads needed. Their tech seamlessly integrates smart assistants into iMessage and SMS, turning your regular texts into powerful AI tools. With monthly messages soaring past 30 million, Linq's solving our app fatigue while keeping conversations flowing naturally.

Microsoft's massive AI investments hit a snag as only 3.3% of users convert to paying Copilot customers. While boasting 15 million paid seats, this pales against their 450 million commercial user base. With $37.5 billion spent last quarter on AI development, executives defend the long-term strategy even as internal reviews may trim underused features.

China's mobile internet landscape is buzzing with AI excitement. Recent data reveals AIGC apps added over 200 million monthly active users in December 2025, growing at a staggering 150% year-over-year. Meanwhile, AI plugins reached nearly 700 million users as smart tools become indispensable for work and daily life. This explosive growth signals how deeply artificial intelligence is reshaping digital experiences.

OpenAI's aggressive shift toward commercializing ChatGPT has triggered internal upheaval, with several top researchers departing. The $500 billion AI leader is reallocating resources from projects like Sora and DALL-E to focus on its flagship product, causing tensions between research purists and commercialization advocates. While investors see potential in ChatGPT's massive user base, the company faces challenges retaining talent amid this strategic pivot.

In a humbling revelation, leading AI models including GPT-4o scored dismally on a rigorous new test designed by global experts. The 'Ultimate Human Exam' exposed critical limitations in AI reasoning, with top performers barely scraping 8% accuracy. These results challenge our assumptions about artificial intelligence's true capabilities and raise questions about whether current benchmarks measure real understanding or just sophisticated pattern matching.

A former OpenAI researcher reveals a critical flaw in today's AI systems: they can't learn from mistakes. Jerry Tworek, who helped develop key models at OpenAI, explains why this inability to adapt threatens progress toward true artificial general intelligence. Unlike humans who evolve through trial and error, current AI hits a wall when facing unfamiliar challenges - forcing experts to rethink fundamental architectures.

Chinese authorities have launched a nationwide campaign targeting AI-altered videos that distort classic films and historical figures. Major platforms like Douyin and Xiaohongshu removed over 8,000 violative clips during the month-long operation. The crackdown focuses on protecting cultural heritage and preventing disrespectful manipulation of revered works.

As businesses adopt multiple AI tools, managing different models has become chaotic. Enter the Large Model Gateway - a traffic cop for AI systems that simplifies access while cutting costs. One company slashed expenses by creating a centralized 'model marketplace' with unified APIs. This innovation could reshape how enterprises deploy AI across departments.

WeChat's Mini Program Education Platform has become a game-changer for schools embracing AI education. With over 170,000 student projects created across 5,000+ global schools, the platform offers free coding tools and AI teaching assistants. Features like natural language programming help students bring ideas to life - like one Guangdong school blending AI with traditional culture studies.

Anthropic is set to release Claude Sonnet5 (codenamed Fennec), a game-changing AI programming model that outperforms its flagship while costing half as much. With a record-breaking SWE-Bench score of 80.9% and innovative 'swarm' development capabilities, this model can autonomously handle entire software projects. The timing appears strategic, coming just as OpenAI prepares its Codex suite launch.

A surprise leak has tech circles buzzing about Anthropic's upcoming Claude Sonnet5. This mid-range AI model appears to outperform its flagship sibling while costing half as much. Its standout feature? The ability to split into specialized 'team members' handling different coding tasks simultaneously. With competitive pricing and groundbreaking swarm intelligence, Claude5 could reshape how developers work.

As Lunar New Year approaches, China's AI leaders are preparing major model releases that could reshape the industry landscape. Zhipu AI's GLM-5 promises breakthroughs in creative tasks, while MiniMax's M2.2 targets developers with enhanced coding capabilities. Meanwhile, DeepSeek takes a more cautious approach with minor updates, saving its firepower for a trillion-parameter model later this year. The holiday season is shaping up to be a pivotal moment in China's AI race.

OpenAI has unveiled its Codex application for macOS, powered by the GPT-5.2-Codex model. While it leads in some benchmarks, competitors like Gemini 3 and Claude Opus remain close behind. The app introduces features like automated task scheduling and flexible agent interaction modes, aiming to streamline developers' workflows. CEO Sam Altman highlights its potential to simplify complex coding tasks, though user experience improvements may still be needed.

Adobe shocks animators by announcing the sunset of its veteran animation tool Adobe Animate in 2026, marking a strategic pivot toward AI. While enterprise clients get extended support until 2029, the creative community voices frustration over the lack of direct alternatives, forcing many to explore competing platforms.

Seattle-based Carbon Robotics has launched its Large Plant Model (LPM), an AI system trained on millions of plant images that can instantly recognize and eliminate weeds without prior exposure. The breakthrough technology eliminates the need for lengthy retraining periods, allowing farmers to specify crops and weeds through a simple interface. This innovation promises to revolutionize weed control efficiency across global farmlands.

Waymo has secured a massive $16 billion investment, sending its valuation soaring past $100 billion. The self-driving pioneer plans to deploy thousands more autonomous vehicles across 20 new cities worldwide by 2026. With Dragoneer leading the funding round, industry watchers see this as Waymo's final push before going public.

Moonshot AI's latest K2.5 model is making waves internationally, with overseas revenue now surpassing domestic income for the first time. The upgraded AI system has seen its global paid user base quadruple, climbing to third place on Openrouter's popularity rankings. Key innovations like native multimodal architecture and 'Agent Swarm' technology are driving this expansion, transforming Kimi from a simple chatbot into a powerful productivity tool.

Chinese optics firm Lingyun is making strategic moves to position itself in the growing AI vision sector. The company plans to cancel 3.48 million shares originally meant for employee ownership while benefiting from China's new 'AI + Manufacturing' policy push. Their high-precision imaging technology is already gaining traction internationally.

SpaceX has filed an audacious plan with the FCC to launch over a million satellites, not for internet service but to create orbiting AI data centers. The proposed space-based supercomputers would leverage the vacuum and cold of space to achieve unprecedented computing power while solving Earth-bound data centers' biggest headaches: energy use and heat dissipation. While promising revolutionary computing capabilities by 2030, the plan faces daunting technical and regulatory hurdles that could ground Elon Musk's latest sky-high ambition.