Microsoft AI Model Development Exposes 38TB of Sensitive Data

Translated data: Microsoft recently experienced a data breach on GitHub, resulting in the leakage of 38TB of private data, including passwords, keys, and internal messages, due to the misuse of shared access signature tokens (SAS) on the Azure platform. The improper use of SAS tokens and lack of monitoring exposed the data for years, highlighting the security challenges in AI model training. This incident underscores the need for stronger security measures and collaborative efforts in the AI development process that relies heavily on large-scale data.

Canva Launches a New Creative Operating System, Fully Upgrading Digital Marketing Tools

Canva launches new digital marketing and video editing tools based on the world's first 'Design AI Model', upgrading its visual suite products, positioning them as a creative operating system for marketing teams. This term does not refer to a traditional operating system, but rather a comprehensive system integrating task tools, AI support, and platform interface.

Microsoft Expands Its Autonomous AI Landscape! After Voice and Chatbots, Microsoft Launches Text-to-Image MAI-Image-1

Microsoft launches its first self-developed text-to-image generator, MAI-Image-1, marking a new stage in its AI development. The model effectively avoids repetition and stylistic issues by incorporating feedback from creative professionals, and it performs exceptionally well in generating photorealistic images, especially excelling in handling complex scenes such as lightning and landscapes.

Anthropic Unveils a Major Update! Claude Sonnet 4.5 Outperforms GPT-5, the New King of Coding

Anthropic released the Claude Sonnet 4.5 model, hailed as the best coding model in the world. The model leads in the SWE-bench coding benchmark test, supports web, mobile applications, and API interfaces, and has demonstrated a sustained operation of 30 hours, achieving significant breakthroughs in handling complex tasks and autonomous agent capabilities.

LiquidAI Launches the Liquid Nanos Series of Lightweight AI Models for Edge Device Applications

LiquidAI launches the lightweight AI model series called Liquid Nanos, specifically designed for edge computing devices such as Raspberry Pi. This series offers two parameter versions, 350M and 1.2B, supporting five application scenarios: translation, information extraction, RAG, tool calling, and mathematical reasoning. It meets the needs of low power consumption and high performance, providing developers with diverse edge AI solutions.

Google Updates Gemini 2.5 Flash Lite to Become the Fastest Proprietary Model

Google updates the Gemini series model, focusing on improving the speed and efficiency of Gemini 2.5 Flash and Flash Lite. Third-party evaluations show that Gemini 2.5 Flash Lite has become the fastest proprietary model, with an output speed of 887 tokens per second, a 40% improvement over the previous generation. This update demonstrates Google's continued commitment to optimizing AI technology.