Google Labs' Portraits experiment allows conversational AI interactions with trusted experts, starting with bestselling author Kim Scott's AI-powered coaching on communication and leadership.
Windsurf, a popular vibe-coding startup reportedly being acquired by OpenAI, says Anthropic significantly reduced its direct access to Claude AI models, forcing it to find alternative providers and potentially causing availability issues for users accessing these models on Windsurf's platform.
SAP announces massive Business AI expansion to over 400 scenarios by 2025, transforming Joule into an AI-powered interface with autonomous workflow agents that can execute complex tasks across multiple applications while eliminating manual prompt engineering.
Anthropic has launched Claude Explains, a blog featuring AI-generated content with human oversight, showcasing its Claude model's writing abilities while emphasizing collaboration between AI and human expertise to create educational resources across diverse topics.
Chinese tech giants and startups are gaining ground in the global race to develop AI agents that can automate complex tasks, with companies like Butterfly Effect, Alibaba, Zhipu, and ByteDance launching agents powered by their own AI models to compete with OpenAI and others.
Google quietly released an Android app called Google AI Edge Gallery that lets users download and run open-source AI models locally on their phones without an internet connection, enabling offline access to capabilities like image generation, question answering, and code writing.
Schools nationwide abandon unreliable AI detection software that falsely accused students of cheating and adopt transparency-based policies requiring students to disclose AI usage while redesigning assignments to promote genuine engagement and learning.
Google's AI assistant Gemini will now automatically summarize long emails, displaying key points at the top, unless users opt out of this feature, marking AI's increasing presence in everyday software and services despite concerns over accuracy.
New Transformer models employ techniques like sparse attention, low-rank factorization, and kernel approximations to overcome quadratic complexity, enabling efficient processing of longer sequences while preserving long-range dependency modeling capability.
Chinese AI startup DeepSeek quietly released an upgraded version of its reasoning model DeepSeek R1, outperforming rivals like Meta and OpenAI, shocking markets and sparking concerns about U.S. tech giants' spending, while showcasing China's AI capabilities despite U.S. export curbs.