Discover the Latest AI News📰

OpenAI introduces 'Instant Checkout' in ChatGPT, enabling users to purchase items directly within the chat interface without browser redirects. Initially available for US users and select merchants, the feature leverages Stripe's payment infrastructure and aims to revolutionize AI-driven commerce.

Anthropic's Claude Sonnet 4.5 AI model has outperformed GPT-5 in coding benchmarks, achieving unprecedented accuracy and autonomy in complex tasks. The update introduces new features like checkpointing and enhanced API tools, positioning it as a leader in AI-assisted development.

DeepSeek has launched its experimental V3.2-exp model featuring a novel sparse attention mechanism that dramatically reduces long-context processing costs. Preliminary tests show up to 50% savings on API calls, potentially revolutionizing AI service economics. The open-weight model is now available on Hugging Face for industry validation.

Anthropic's Claude Code 2.0 introduces terminal upgrades, VS Code integration, and checkpoint features alongside the Sonnet4.5 model, promising significant efficiency gains for developers.

Cambrian has announced full compatibility with DeepSeek's experimental V3.2-Exp model while open-sourcing its vLLM-MLU inference engine. The breakthrough leverages Triton and BangC operators for enhanced efficiency in long-sequence processing, marking significant progress in AI model optimization.

Anthropic has launched Claude Sonnet 4.5, featuring top-tier coding capabilities and an experimental 'Imagine with Claude' interface generator. The model scores 77.2% on SWE-bench, surpassing its predecessor, while the new feature dynamically creates software interfaces from natural language prompts, signaling a shift toward AI-native development paradigms.

Startup Anything secures $11M funding led by Footwork, reaching a $100M valuation. The company aims to revolutionize 'ambient coding' by offering full-stack AI tools that turn prototypes into production-ready apps. Competitors like Lovable and Replit show market demand, but Anything's integrated infrastructure sets it apart.

OpenAI is reportedly developing a TikTok-like social media app powered by its Sora2 video generation model. The platform will feature exclusively AI-generated content, with videos capped at 10 seconds. Unique features include identity verification for personalized avatars and copyright protection measures. This move positions OpenAI against competitors like Google's Veo3 in the AI video generation space.

Former Yahoo CEO Marissa Mayer has closed her consumer software company Sunshine, selling its assets to AI startup Dazzle. Sunshine's team will transition to Dazzle, which aims to develop an intelligent personal assistant. Despite launching multiple products, Sunshine struggled with market adoption due to privacy concerns.

Baidu Maps has unveiled DuerOS 2.0, an upgraded intelligent travel assistant featuring deep integration with a speech language model. The update enhances navigation accuracy, cross-device functionality, and personalized recommendations, marking a significant leap in AI-driven travel services.

OpenAI has launched new parental controls and a security routing system for ChatGPT, aiming to enhance safety for younger users and handle sensitive conversations more effectively. The updates include model switching for sensitive topics and customizable teen account settings, though some users criticize the approach as overly restrictive.

Chinese AI lab DeepSeek, backed by quant fund High-Flyer Capital, has gained global attention with its efficient models outperforming competitors. Despite U.S. chip restrictions, its V3 model rivals GPT-4o while triggering industry price cuts and regulatory scrutiny.

OpenAI is launching Sora2, a TikTok-like app with AI-generated videos, and integrating instant shopping into ChatGPT via partnerships with Etsy and Shopify. The moves mark OpenAI's shift toward consumer applications amid ongoing copyright concerns.

Microsoft introduces 'Vibe Working,' an AI-driven office assistant deeply integrated into Excel, Word, and PowerPoint. Featuring Agent Mode and Office Agent, it aims to automate tasks and streamline workflows. While promising efficiency gains, concerns linger about potential errors.

Former Microsoft executives have founded Maximor, an AI-driven company aiming to reduce reliance on Excel for financial management. The startup has secured $9 million in seed funding to develop AI agents that integrate with ERP and CRM systems, streamlining accounting processes. Early adopters like Rently have cut monthly closing times by 50%. The solution supports both US GAAP and IFRS standards for global enterprises.

Brave Browser has launched 'Ask Brave,' an AI-powered search feature that delivers detailed, report-like answers while prioritizing user privacy. The tool coexists with Brave's existing AI Answers and processes over 15 million queries daily. Unlike competitors, Brave encrypts chat records and deletes them after 24 hours of inactivity.

Microsoft has restructured its Windows engineering team, merging client and server divisions under a single leader to accelerate its AI operating system strategy. The move centralizes control of Windows development and strengthens collaboration with Azure while focusing on AI-powered features like Copilot Vision and the new 'Windows AI Labs' initiative.

Ant Group has open-sourced Ring-1T-preview, the world's first trillion-parameter inference model. The preview version demonstrates superior performance in natural language reasoning and code generation, surpassing competitors like Gemini2.5Pro and approaching GPT-5 capabilities.

Heidelberg, Germany, has launched an AI-powered 'Cityscanner' system to combat illegal parking. The technology uses cameras, lidar, and radar to identify violations with 99.7% accuracy, connecting to municipal databases for real-time verification. The system is seven times more efficient than manual patrols and is projected to generate €1.2 million annually in fines.

YouTube has removed a channel named 'Woman Shot AI' for posting AI-generated videos depicting violent scenes of women being killed. The channel, which gained over 175,000 views and 1,200 subscribers, featured extreme misogynistic and racist content. The platform acted after media scrutiny, citing violations of its service terms. The incident highlights ongoing concerns about AI misuse and the need for stricter regulations.