Google's AI assistant Gemini now analyzes videos in Google Drive, extracting summaries and key insights. Users can request brief overviews, action items, or specific details from meeting recordings or training sessions without rewatching entire files. Accessible via the "Ask Gemini" button, this feature uses captions to process content and is available to select Google Workspace and Google One subscribers. It streamlines information retrieval, saving time for those managing extensive video libra

DeepSeek released a lightweight R1 variant, DeepSeek-R1-0528-Qwen3-8B, built on Alibaba's Qwen3-8B. It outperforms Google's Gemini 2.5 Flash in AIME 2025 and rivals Microsoft's Phi 4 in HMMT math tests. Requiring fewer computational resources than full-sized models, it is licensed under MIT for commercial use and available via APIs on platforms like LM Studio.

Meta AI has reached 1 billion monthly active users, doubling from 500 million in September 2024. CEO Mark Zuckerberg highlights plans to deepen user engagement and expand AI assistant capabilities, including potential premium features and subscription services for additional computing resources, positioning Meta AI to compete with apps like ChatGPT. This milestone comes a month after the launch of the standalone Meta AI app.

Hugging Face unveiled two new open-source humanoid robots, HopeJR and Reachy Mini. HopeJR is a full-sized robot with 66 degrees of freedom for walking and arm manipulation, while Reachy Mini is a compact desktop unit for AI testing. Prices range from $250-$300 for Reachy Mini and around $3000 for HopeJR. Delivery is expected by year-end. Enabled by the acquisition of Pollen Robotics, these affordable, open-source robots aim to democratize robotics. Hugging Face has been expanding in robotics sin

Google's Gemini AI can now summarize videos in Google Drive, extract action items, and answer questions about the content. Accessible via Drive's overlay preview or a new browser tab, this feature targets the "meeting recording paradox" by offering instant summaries and insights without requiring rewatching. It uses captions and transcripts to analyze videos, supporting only English for now. Available to Google Workspace and Google One AI premium users, it aims to improve workflow efficiency and

Claude AI by Anthropic introduces a beta conversational voice mode, available in English on iOS and Android. This feature allows real-time audio interactions, with key points displayed on screen for clarity. Users can switch between text and voice input seamlessly. Voice-driven access to connected services like Google Docs and Gmail is exclusive to paid subscribers. Pricing tiers include Free, Pro ($17/$20 per month), and Max ($100+ per user per month). Anthropic recommends using the feature in

Nvidia and AMD plan to launch new AI GPUs for the Chinese market to comply with US export restrictions. Nvidia's "B20" and AMD's Radeon AI PRO R9700 are expected to debut in July. Nvidia also develops a cost-effective Blackwell-based chip priced at $6,500-$8,000. Restrictions caused $4.5 billion in charges and $2.5 billion in lost sales for Nvidia in Q1, with an estimated $8 billion Q2 revenue impact.

Akool, a generative AI company, launched Akool Live Camera, a tool for real-time AI-generated videos. It creates lifelike avatars that mimic voice, gestures, and expressions for live interactions. Unlike pre-recorded tools, it focuses on live streaming, virtual meetings, and AR gaming with features like multilingual translation and AI dubbing. Backed by investors including G2C Venture Partners, Akool has raised $40 million in funding.

Microsoft launched the beta of its gaming assistant, Copilot, via the Xbox app on iOS and Android. It answers gaming-related questions using users' account data and gaming history, while integrating Bing's public info. The feature is available in English for users aged 18+ in select regions, with plans to expand globally. Future updates will enhance personalization and proactive guidance.

Google is revamping its Photos editor with AI-powered features like Reimagine and Auto Frame, previously exclusive to Pixel devices. Reimagine uses generative AI to alter photos via text prompts, such as changing a dull sky to a clear blue one. Auto Frame suggests alternative compositions using AI to fill gaps. The new editor consolidates tools in one place, offering AI-driven suggestions for quick edits. Tapping specific image areas provides tailored recommendations, like adjusting lighting or

DeepSeek's R1 model update showcases improvements in reasoning, output quality, and user-recognized enhancements. Despite slower response times, the AI demonstrates stronger inference capabilities and coding performance, positioning DeepSeek as a competitive, affordable alternative in China's growing influence on the global AI landscape.

Jules, Google's AI-powered coding assistant, offers automated code updates, bug fixes, and GitHub integration using Gemini 2.5 Pro. In public beta since May 2025, it performs tasks like test creation, feature development, and dependency updates asynchronously. Free during the beta with five daily requests, Jules aims to enhance developer workflows by analyzing codebases, generating audio summaries, and providing reviewed change suggestions.

Anthropic is beta testing Voice Mode for its Claude chatbot, enabling spoken conversations in English. Powered by Claude Sonnet 4, it allows hands-free interactions with voice responses and on-screen key points. Users can choose from five voice options, switch between text and voice, and access transcripts. Free users have limited conversations, while paid subscribers enjoy additional features like Google Workspace integration. Voice Mode competes with similar offerings from OpenAI, Google, and

OpenAI is exploring a "Log in with ChatGPT" feature to allow users to access third-party apps via their ChatGPT accounts. With 600 million monthly users, OpenAI aims to expand into domains like online shopping and social media. The feature is currently in testing with Codex CLI, offering API credits to Plus and Pro users. OpenAI seeks developer interest for integration, targeting apps of all sizes. Though previously hinted at by CEO Sam Altman, details on the launch timeline remain unclear.

Meta is reorganizing its AI division into two teams: AI Products, focusing on consumer features for Facebook, Instagram, WhatsApp, and standalone AI apps; and AGI Foundations, working on advancing Llama models. This restructuring aims to boost Meta's competitiveness against OpenAI and Google without layoffs.

Google DeepMind's AlphaEvolve uses large language models to discover and optimize algorithms in various domains. It has improved Google's data center operations, AI training processes, and rediscovered state-of-the-art solutions for mathematical problems. Despite its potential impact on fields like materials science and drug discovery, there are concerns about DeepMind's lack of full open-sourcing, as seen with previous projects like AlphaFold2 and AlphaGeometry. Academic researchers can apply f

Claude 4 Opus, developed by Anthropic, excels in complex coding, research, and agent-driven workflows. It offers hybrid reasoning for instant or in-depth responses, industry-leading coding capabilities with 72.5% on SWE-bench, and robust agent workflows for enterprise tasks. The model supports up to 32,000 tokens, integrates with IDEs, and provides secure ASL-3 safety protocols. Pricing starts at $15 per million input tokens, with API access via Anthropic, Amazon Bedrock, and Google Cloud.

Cloudflare launched 13 new MCP servers to enhance AI agent integration with its platform. These servers enable natural language interactions for tasks like debugging, data analysis, and security monitoring. Key features include Workers Observability for logs, Radar for traffic data, Logpush for log management, AI Gateway for model inspection, AutoRAG for document retrieval, DNS management, KV storage access, Pages deployment control, Queue task handling, R2 object storage, Turnstile CAPTCHA mana

Duolingo CEO Luis von Ahn reassures employees and the public about the company's AI-first strategy, acknowledging prior unclear communication. He emphasizes that AI aims to enhance productivity without replacing human workers, introduces adaptation measures like workshops, and addresses fears of job displacement. The shift follows controversy over phasing out contractors for AI-driven tasks, part of broader efficiency efforts. While Duolingo joins companies like Shopify and Klarna in embracing A

Elon Musk announces a return to a 24/7 work schedule across his companies, aiming to refocus on Tesla, SpaceX, xAI, and X after criticism of divided attention. Amid concerns over political involvement and its impact on businesses, Tesla's Q1 2025 results showed a 71% net income drop and 9% revenue decline. Recent global outages on X highlight infrastructure reliability issues, prompting Musk to prioritize improvements and technological advancements.