Google Classroom introduces an AI-powered quiz generator using Gemini AI, allowing teachers to create quizzes from uploaded files or text. It analyzes content, generates customizable questions aligned with teaching goals, and exports quizzes to Google Docs or Forms. This feature is available to Gemini Education subscribers and aims to enhance global teaching efficiency.

Nvidia is moving part of its chip production to the US, manufacturing Blackwell series AI GPUs at TSMC's Arizona facility. This marks Nvidia's first advanced AI chip production line in the US. TSMC is investing $100 billion in US expansion, with its Arizona plant producing 4-nanometer chips and targeting 2-nanometer technology by 2030. Nvidia also partners with Foxconn and Wistron for supercomputer production in Texas, aiming to boost output over 12-15 months. This strategic shift is expected to

OpenAI launches GPT-4.1, an upgraded model with enhanced coding and instruction-following capabilities, alongside lightweight versions GPT-4.1 Mini and Nano. These models support up to one million tokens, offer a 26% cost reduction, and will replace GPT-4. GPT-4.5 API access ends on July 14. The release confirms delayed GPT-5 plans and hints at upcoming o3 and o4 mini inference models.

Apple unveiled a new AI initiative to enhance model performance while safeguarding privacy. Devices will use synthetic datasets and user-shared samples for training, comparing them locally without transmitting raw data. Only signals indicating the most similar synthetic samples are sent to Apple's servers. This system, integrated into beta versions of iOS, iPadOS, and macOS, leverages differential privacy technology to ensure data anonymity. The move aims to balance improved AI functionality wit

Google is pushing into AI at Cloud Next, unveiling a new AI processor and services for autonomous AI capabilities. Despite tariff concerns impacting tech stocks, funding for AI continues to grow with A16z raising $20 billion and Signalfire securing $1 billion for AI startups. Musk's DOGE initiative also plans a hackathon for IRS software.

Google Cloud announced major updates to its data analytics platform at the 2025 Cloud Next conference, integrating Gemini into BigQuery and Looker for enhanced code assistance, analytics, and specialized agents. The new BigQuery Knowledge Engine improves context understanding, boosting accuracy and productivity. These autonomous features aim to reduce manual work by suggesting connections and recommending transformations, with impressive results like a 60% acceptance rate for SQL/Python assistan

OpenAI plans to replace GPT-4 with GPT-4o in ChatGPT by April 30, 2025. GPT-4o, launched in May 2024, integrates multimodal capabilities including text, visual, and audio inputs. It offers real-time responses, emotional voice interactions, and supports multiple languages. The model is free for all users with enhanced limits for ChatGPT Plus subscribers. Despite controversies like the "Sky" voice resembling Scarlett Johansson's, GPT-4o marks a major advancement. OpenAI also countersued Elon Musk

Anthropic published two papers exploring the inner workings of large language models, introducing an 'AI Microscope' to identify interpretable concepts and their computational circuits. The research examines behaviors like planning, hallucinations, and multilingual capabilities, revealing models use a universal language for concept generation. It also highlights mechanisms behind anti-hallucination and reasoning processes, aiming to improve model interpretability and alignment with human values.

Safe Superintelligence (SSI), an AI startup led by Ilya Sutskever, former OpenAI chief scientist, has raised $20 billion in new funding, valuing it at $32 billion. Previously securing $1 billion, SSI focuses on developing safe superintelligence, though its product remains under development. The round was reportedly led by Greenoaks, but SSI has not officially commented.

OpenAI may soon implement an identity verification process for organizations accessing advanced AI models. Dubbed "Verify Organization," this procedure aims to enhance security and prevent misuse, requiring a government-issued ID from supported countries. Not all organizations will qualify, and each ID can verify only one organization every 90 days. This move follows reports of policy violations, including alleged data extraction by a China-linked group in late 2024, and seeks to balance accessi

Netflix is testing an OpenAI-powered search feature allowing users to apply specific criteria, including emotional state. Currently in a voluntary trial on iOS in Australia and New Zealand, it will expand to the US. The feature, still in early stages, aims to broaden search scopes beyond genres or actor names. Netflix has used AI in recommendations for over two decades. OpenAI has not yet commented on the collaboration.

ByteDance is developing AI-powered smart glasses, expanding its presence in the wearable device market. The product may feature voice assistants, real-time translation, and health monitoring, competing with Meta's Ray-Ban smart glasses. While details remain undisclosed, ByteDance aims to combine style and functionality, enhancing its competitive edge. Release timeline and sales regions are yet to be announced.

LinkedIn users are embracing the "toy-fication" trend, transforming themselves into AI-generated plastic toy figures with professional accessories. While gaining traction, it remains less popular than the Studio Ghibli art wave. ChatGPT's image generator, despite usage limits imposed by OpenAI, plays a key role in this emerging trend, mostly shared by marketers on LinkedIn with limited engagement.

Apple plans to launch an updated Siri in fall 2025 with smarter capabilities, but delays and internal management issues raise uncertainty. Leadership changes, resource allocation conflicts, and hardware limitations in AI chip development pose significant challenges. Observers question if Apple can meet its deadlines amid these obstacles.

LiveKit, an open-source software for real-time audio and video streaming, has evolved into a commercial success with over 500 paying customers and 100,000 developers using its platform. Founded by Russ d’Sa and David Zhao, it supports major companies like Spotify and Meta, powers 25% of US 911 calls, and integrates with OpenAI's ChatGPT voice mode. Recently, LiveKit raised $45 million in Series B funding and is developing scalable AI-driven communication solutions, positioning itself as an AI-na

Canva introduces Canva AI, featuring an AI assistant for design tasks, image editing tools, and app creation via prompts. New tools like Canva Code enable building interactive applications, while Magic Insights and Magic Charts enhance enterprise data visualization. Despite concerns about AI's impact on creativity and jobs, Canva emphasizes AI as a growth driver for various roles. Recent layoffs were unrelated to AI development but part of company restructuring.

Google plans to integrate its Gemini AI model with the Veo video generation model to enhance physical world understanding. Designed as a multimodal foundation, Gemini aims to become a universal digital assistant capable of handling real-world scenarios. Veo's training data largely comes from YouTube, enabling it to learn physical world rules. The AI industry is moving toward all-in-one models requiring massive multimedia training datasets. Google has expanded terms of service for AI system train

OpenAI enhances ChatGPT with improved memory, allowing it to reference past conversations for more personalized interactions. Available to Plus and Pro users globally (except UK/EU), the feature ensures user control over privacy, offering options to disable or temporarily opt-out of memory use. This update aligns with efforts by AI firms to create persistent digital companions.

Google collaborates with Range Media on "AI on Screen," a short film project exploring the complex human-AI relationship. Through cinematic storytelling, it highlights AI's opportunities, challenges, and ethical dilemmas, offering a new perspective on the technology's impact. Two films are in production, set for release later this year.

Microsoft is slowing or pausing some early-stage data center projects to optimize its construction strategy amid rising costs. This move, described as part of normal business management, focuses on feasibility studies requiring 1-3 years of development. It won't affect AI investments, with Azure continuing to support over a million AI servers. The adjustment aims to align with tech iterations and market demands while exploring innovations like energy-saving and liquid cooling solutions.