Tech Roundup: Anthropic Claude for Word, Google Gemini Notebooks & More

[A recurring feature on the latest in Science & Technology.]

• OpenAI publishes a five-principle framework for AGI development, pledging to resist concentrating AI power and to collaborate with companies and governments.
• Anthropic says Opus 4.7 is "less broadly capable" than Claude Mythos Preview and that its "cyber capabilities are not as advanced as those of Mythos Preview"; details using AI agents to accelerate alignment research on "weak-to-strong supervision," where a weak model supervises the training of a stronger one.
• Spotify debuts a new look for listeners on tablets, offering an adaptive orientation that switches between portrait and landscape layouts, as well as a collapsible sidebar; expands Prompted Playlists, its AI feature for creating playlists, to include podcasts, for Premium users in the U.S., Canada, the U.K. and other countries.
• Opera adds Browser Connector for integrating AI chatbots, allowing Opera One and Opera GX users to integrate either ChatGPT or Claude into the platform.
• A U.S. jury finds that Live Nation and Ticketmaster illegally maintained monopoly power in the ticketing market, in a case brought by state attorneys general after the federal government struck a settlement with Live Nation in March 2026.
• Google rolls out Gemini 3.1 Flash TTS, a text-to-speech model with support for over 70 languages and audio tags that give developers granular speech control.
• Meta researchers introduce "hyperagents" to unlock self-improving AI for non-coding tasks, such as robotics and document review.
• Handset maker Nothing makes it easy to share files between any Android phone and a Mac using a new app called Warp.
• Amazon launches the latest version of Fire TV Stick HD with Alexa+ built in and a redesigned streaming experience.
• Facebook, Instagram, YouTube and TikTok now capture over 90% of social media ad revenue, with Meta accounting for 70% of the total.
• Anthropic opposes an Illinois bill backed by OpenAI that would shield AI labs from liability, even for "critical harms" like more than 100 deaths or $1 billion in damages, if they published safety reports.
• The FCC grants Netgear a conditional approval to import its future consumer routers, cable modems and cable gateways into the U.S. through October 1, 2027.
• Google says YouTube livestreams will hold back ads during peak engagement to protect the vibe, as well as show fewer ads to users who support creators with Super Chat, Super Sticker or gift purchases.
• More than 70 civil society organisations call on Meta to "immediately halt and publicly disavow" its plans to deploy facial recognition features on its Ray-Ban and Oakley glasses, including a feature known as "Name Tag" that would reportedly allow users to identify people.
• Roblox unveils Kids accounts for ages 5-8 and Select accounts for ages 9-15, with age verification features; says developers will need Roblox Plus, a new $4.99-per-month subscription offering benefits like discounts, to publish games for Kids and Select accounts.
• OpenAI buys AI personal finance startup Hiro for an undisclosed sum, as it aims to build financial planning capabilities into ChatGPT.
• Microsoft says it is "exploring the potential of technologies like OpenClaw in an enterprise context", including a team of always-on agents within Microsoft 365.
• Google rolls out a dedicated Google Meet app for Android Auto after debut a similar feature for CarPlay at the beginning of April; unveils speech translation feature within Meet on Android to facilitate bidirectional translation between English and Spanish, French, German, Portuguese and Italian.
• X says it's cutting back on payments to accounts that are "flooding the timeline" with clickbait and rapid-fire news aggregation.
• Amazon Luna drops support for game purchases and third-party game stores and subscriptions; previously purchased games will be accessible until June 10, 2026.
• Google updates Gemini with the ability to generate interactive 3D models and simulations. (Users must select the Pro model in the prompt bar.)
• Meta's Instagram today implements a small but useful change that allows users to edit their comments for up to 15 minutes after writing the initial comment.
• Anthropic makes Claude Cowork, previously available as a "research preview," generally available to all paid plans, and adds six features for enterprise use; announces Claude Managed Agents, offering developers an agent harness and other tools to build and deploy AI agents at scale.
• X brings back Voice Notes to X Chat, supporting one-on-one and group conversations via a new voice input icon; rolls out automatic post translation worldwide, powered by xAI's Grok, and updates its iOS image editor with features like drawing, text tools and blurring.
• Spotify adds toggles to stop video from playing inside the app for both music and podcasts on all platforms and devices.
• Tubi becomes the first streamer to launch a native app within ChatGPT, allowing viewers to find movies or shows to watch by using conversational phrases.
• Meta releases Muse Spark, the first model from Meta Superintelligence Labs under Alexandr Wang, to "power a smarter and faster" Meta AI across Meta's products; says Muse Spark powers queries in Meta AI and its "shopping mode" feature and that it plans to release a version of Muse Spark under an open-source license.
• Google introduces "notebooks" in the Gemini app for deeper NotebookLM integration and a space to organize chats and files, after adding sources in December 2025.
• AWS debuts Amazon S3 Files, a new capability built on Amazon's Elastic File System that lets applications and AI agents access S3 buckets as local file systems.
• Anthropic signs an agreement with Google and Broadcom for multiple GWs of TPU capacity.
• Adobe launches Acrobat Spaces, a free AI-based study tool to help students generate flashcards, mind maps, quizzes and Express presentations without logging in.
• Netflix launches Netflix Playground, a standalone games app for kids aged eight and under, in the U.S., the U.K., Canada, Australia, the Philippines and New Zealand.
• ByteDance launches its Seedance 2.0 video model to enterprise clients in 100+ countries, excluding the U.S. amid legal disputes, after a February launch in China.
• A U.S. judge issues a $322.2 million judgment against pirate library Anna's Archive for scraping Spotify, a largely symbolic victory as the site is anonymously operated.
• Microsoft debuts MAI-Image-2-Efficient, a faster version of its flagship text-to-image model, which it says offers production-ready quality at ~50% the cost.
• Amazon unveils the Amazon Leo Aviation Antenna, saying it can deliver up to 1 Gbps download and 400 Mbps upload speeds "simultaneously" for in-flight Wi-Fi.
• Google and Intel expand their partnership to deploy Xeon chips, including Xeon 6, and co-develop custom Infrastructure Processing Units to improve efficiency.
• Visa unveils Intelligent Commerce Connect, a platform that facilitates payments for AI agents across multiple card networks, including those of Visa competitors.
• OpenAI launches a $100/month ChatGPT Pro subscription, which offers 5x more Codex usage than Plus; the $200/month Pro plan offers 20x higher limits than Plus.
• Microsoft unveils plans to streamline the Windows Insider program, offering a simplified channel structure and the ability to move between channels without wiping PCs and get access to experimental features without having to download a third-party app like ViVeTool.
• France says it plans to move government computers running Windows to Linux, to further reduce its reliance on U.S. technology, without providing a timeline.
• Anthropic debuts Claude for Word in beta, adding AI editing tools and clickable citations, targeting document-heavy workflows, for Team and Enterprise users.
• Meta begins removing dozens of Facebook and Instagram ads from lawyers seeking clients who claim to have been harmed by social media while under the age of 18.
• Google's YouTube launches a Shorts feature to let creators generate photorealistic AI avatars using a "live selfie" recording of their face and voice that's powered by Veo.
• Patreon says it now has 7.6 million paid podcast memberships.
• OpenAI releases the Child Safety Blueprint tackling AI-enabled child sexual exploitation, focusing on updating legislation and improving detection and reporting.
• Amazon says Kindle and Kindle Fire devices released in 2012 and earlier won't be able to access the Kindle Store from May 20, 2026. (However, downloaded books can still be read.)
• Greek Prime Minister Kyriakos Mitsotakis says the country will ban social media access for kids under 15 from January 1, 2027, and calls for coordinated E.U. action.
• OpenAI, Anthropic and Google are sharing information via the Frontier Model Forum to detect adversarial distillation attempts that violate their Terms of Service, according to Bloomberg.
• Google is updating Gemini to add a UI that triggers support hotline referrals and a "help is available" module when chats indicate potential crises like suicide.
• OpenAI announces a Safety Fellowship program for external researchers, engineers, and practitioners to study the safety and alignment of advanced AI systems.