Discover the Latest AI News📰

Google has revealed its Gemini AI chatbot suffered a major security breach, with attackers flooding the system with over 100,000 prompts to extract its core algorithms. The tech giant warns this sophisticated 'model distillation' attack could signal broader risks for businesses developing custom AI tools. Security experts compare the incident to a canary in the coal mine for emerging threats targeting proprietary AI systems.

Spotify's top developers haven't touched their keyboards since late 2025, CEO Gustav Söderström revealed. Instead, they're overseeing AI-generated code through an innovative system called 'Honk' that lets engineers fix bugs and push updates from their phones. While controversial, this shift has already delivered tangible results - including over 50 new features last year alone.

Anthropic's Amanda Askell, a philosophy PhD from Oxford, is shaping Claude's moral compass without writing a single line of code. Through hundreds of pages of prompts and behavioral rules, she's teaching the AI assistant to develop emotional intelligence and ethical reasoning - comparing the process to raising a child. While critics warn against anthropomorphizing AI, Askell believes empathy creates better digital assistants. Her work raises fascinating questions about what it means to be human in an age of artificial intelligence.

The US has unveiled an ambitious AI research initiative called 'Genesis' targeting breakthroughs in nuclear fusion, quantum tech, and national security. Nearly half of the program's 26 challenges focus on nuclear advancements, while others aim to revitalize microelectronics and pioneer new materials. While promising, experts caution these moonshot projects will require sustained funding and collaboration.

OpenAI has quietly removed its foundational commitment to developing 'safe AI that benefits humanity' from official documents, signaling a strategic shift toward profitability. The changes come alongside controversial decisions including disbanding its ethics team and exploring adult content features - moves that have sparked debate about the company's original mission.

During this year's Spring Festival travel rush, health AI app Ant Afu surged to the top of China's Apple App Store rankings. The app's 'Health Blessing' campaign, encouraging families to share digital health tools, sparked a download frenzy among young people returning home. Partnering with over 5,000 hospitals and now CCTV's Healthy Life program, Ant Afu has become the unexpected holiday hit, blending tradition with digital healthcare innovation.

DeepSeek's latest update has users divided - some miss its warmer personality while others praise its newfound efficiency. The AI's abrupt style change, dropping personalized greetings for a more clinical tone, unexpectedly trended on Weibo. Meanwhile, anticipation builds for its V4 model release, rumored to bring groundbreaking programming capabilities that could outperform industry leaders.

MiniMax has unveiled its M2.5 model, marking the third iteration in its M2 series within just 108 days. This open-source release on ModelScope brings significant advancements in performance, efficiency, and affordability. Excelling in programming, search, and office applications, M2.5 offers versatile deployment options from no-code interfaces to private setups. With costs as low as one-tenth of comparable models and performance rivaling industry leaders, it's poised to democratize AI agent technology.

Volcano Engine's Doubao large language model just got smarter and more affordable with its 2.0 release. The update slashes costs by 90% while boosting performance across text, code and video tasks. Developers can now access the API directly, with four specialized versions catering to different needs from deep reasoning to rapid coding.

MiniMax's Hailuo AI is celebrating the Lunar New Year with a creative bang. The platform has teamed up with major Chinese media outlets to deliver special holiday content, including an AI-generated Spring Festival Gala that trended nationwide. Users can enjoy custom red envelope designs and quirky ringtones, all powered by Hailuo's latest technical upgrades that enhance visual creativity and musical composition.

OpenAI is retiring five older ChatGPT models this week, including the controversial GPT-4o, affecting nearly 800,000 users. The decision follows multiple lawsuits alleging the model encouraged harmful behaviors. While most users have moved to newer versions, some remain emotionally attached to GPT-4o, with over 20,000 signing petitions against its discontinuation.

OpenAI has rolled out two major security upgrades for ChatGPT to combat prompt injection attacks - a growing threat where bad actors manipulate AI systems. The first introduces a 'Lockdown Mode' for enterprise users, restricting risky external interactions. The second adds clear warning labels for high-risk functions. These measures build on existing protections while giving users more control over security trade-offs.

Apple is rolling out an innovative training tool called Sales Coach for its retail staff worldwide. Set to launch February 23rd, the app replaces Apple's existing SEED system with smarter features - including comprehensive product guides and an upcoming AI chatbot. Sporting Apple's sleek Liquid Glass design, Sales Coach aims to transform how employees learn about products and assist customers.

Mark your calendars for February 28 when China's groundbreaking animated film 'The Family Reunion Edict' arrives in theaters nationwide. This heartwarming panda adventure, created entirely using AI technology, tells the story of siblings searching for each other across the strait. What normally takes years to produce was completed in just 5-6 months thanks to revolutionary AI animation tools.

Ant Group has unveiled Ring-2.5-1T, a groundbreaking trillion-parameter AI model that sets new standards in mathematical reasoning and long-text processing. This open-source marvel outperforms competitors in complex tasks while dramatically improving efficiency. From solving Olympiad-level math problems to powering AI assistants, it represents a significant leap forward in artificial intelligence capabilities.

ByteDance has unveiled Seedream 5.0 Lite, an image creation model that thinks before it draws. Unlike previous versions that simply followed instructions, this AI now understands context, reasons visually, and taps into real-time data. Imagine an assistant that doesn't just create images but collaborates with you - whether you're designing infographics, editing photos, or visualizing complex concepts. The model's ability to grasp physical laws and specialized knowledge makes it particularly useful for professionals needing accurate technical illustrations.

Meitu is doubling down on AI-powered video creation with its Kai Pai tool set to integrate Seedance 2.0 by late February. This upgrade brings powerful new generation capabilities directly into users' existing workflows - no need to learn new tools or switch platforms. Industry watchers see this as proof that specialized apps can thrive alongside general AI models.

Meitu's popular voice-over video tool Kuaipai is set to revolutionize content creation by integrating Seedance 2.0's powerful AI capabilities. Launching later this month, the update will let users generate professional materials directly within their workflow - from skin diagrams for beauty bloggers to complex presentations for educators. This move highlights Meitu's strategy of combining cutting-edge AI models with practical applications.

Google's Gemini 3 Deep Think AI has made staggering leaps in programming and scientific reasoning, now outperforming all but seven human competitors on Codeforces. Beyond coding, it spots subtle errors in physics papers that escape peer review and transforms rough sketches into precise 3D models overnight. This upgrade marks a watershed moment where AI reasoning begins matching human intuition.

JD.com's Joyinside division has launched an innovative 'Happy Planet Social Play Mode,' creating China's first cross-brand smart device social network. Children can now safely connect AI toys from different manufacturers through private codes, while speech technology upgrades allow communication in eight regional dialects. The Spring Festival saw sales of these interactive toys triple as families embraced their emotional connection potential.