At Google I/O 2025, Google unveiled Veo 3, its most advanced video-generating AI model yet — and this time, it’s not just about visuals. Veo 3 marks a major leap forward by introducing synchronized audio generation, allowing creators to produce fully soundtracked videos from a single prompt. According to Demis Hassabis, CEO of Google DeepMind, Veo 3 can now add dialogue, sound effects, and ambient noise to match the video scenes it creates. “For the first time, we’re emerging from the silent era of video generation,” said Hassabis during the launch event. Key Features: Veo 3 builds on the foundations laid by earlier research from DeepMind into video-to-audio AI. While Google hasn’t disclosed the specific datasets used, YouTube is believed to be a major contributor due to its vast video library and Google’s ownership of the platform. Updates to Veo 2 In parallel, Google has upgraded Veo 2 with features aimed at improving creative control: These enhancements will be rolled out soon via Google’s Vertex AI API, making Veo more accessible to developers and content creators. While Veo 3 promises groundbreaking creative power, it also raises industry concerns. A 2024 study by the Animation Guild warned that over 100,000 media jobs in the U.S. could be disrupted by AI by 2026. As Google continues advancing these technologies, balancing innovation with ethical deployment remains critical. Source: (Techcrunch)
Google I/O 2025: Every Major AI, Android & Hardware Announcements
Google’s biggest developer event of the year, Google I/O 2025, delivered a wave of innovation across AI, Android, hardware, and developer tools. Held at the Shoreline Amphitheatre in Mountain View, the event featured updates that reinforce Google’s AI-first direction while enhancing user experience across devices and services. Gemini Ultra & Gemini 2.5 Pro “Deep Think” Google I/O introduced Gemini Ultra, a $249.99/month premium AI subscription that grants access to powerful tools like Veo 3 (AI video generator), Imagen 4 (AI image generator), and the advanced Deep Think mode in Gemini 2.5 Pro. Deep Think enables more nuanced reasoning by evaluating multiple answer paths before responding. It’s available to trusted testers via the Gemini API. AI Content Tools: Veo 3, Imagen 4, & Flow Veo 3 now generates sound effects and dialogue along with video, while Imagen 4 offers faster, more detailed image creation. Both tools are integrated into Flow, Google’s AI-driven video editing suite. Gemini App Expansions The Gemini app now supports Live Camera and Screen Sharing on iOS and Android, allowing near-real-time interaction powered by Project Astra. Deep integration with apps like Maps, Calendar, and Tasks is also rolling out. Gemini’s Deep Research now supports uploading private documents for analysis. Stitch & Developer AI Tools Google launched Stitch, an AI UI builder for generating HTML and CSS from text or image prompts. Developers also gained wider access to Jules, an AI assistant for debugging, GitHub pull requests, and backlog management. Project Mariner & Project Astra Project Mariner, Google’s AI that browses websites and performs tasks on behalf of users, now supports multitasking. Project Astra, developed by DeepMind, enables near-instant multimodal AI experiences and is being used in Gemini, Search, and smart glasses being co-developed with Samsung and Warby Parker. AI Mode in Search AI Mode, a new Google Search feature, allows users to ask complex, multi-part questions. A “Search Live” feature is coming soon to let users ask questions based on their phone’s camera view in real time. Beam 3D Teleconferencing Google’s Beam (formerly Project Starline) provides life-like 3D video calls using AI-powered rendering and real-time speech translation. Beam is integrated with Google Meet for natural, immersive communication. More AI & Workspace Features NotebookLM & SynthID NotebookLM introduces Video Overviews, turning source material into brief AI-generated videos. Google also rolled out SynthID Detector, a tool to verify AI-generated content. Wear OS 6 & Pixel Watch Wear OS 6 brings a refreshed UI with dynamic theming and unified fonts, enhancing personalization across the Pixel Watch series. Google Play Upgrades The Play Store now features topic pages, audio previews, and improved subscription management with multi-product checkout. Developers also gain enhanced release management tools and app monitoring. Android Studio AI Upgrades Android Studio adds Agent Mode and Journeys, new AI-powered development assistants, and upgrades to crash detection using Gemini. Source: (Techcrunch)
Google Partners with Warby Parker to Launch AI Smart Glasses
In a major move unveiled at Google I/O 2025, Google announced a strategic partnership with Warby Parker to develop a new generation of AI-powered smart glasses. Backed by a commitment of up to $150 million, the collaboration aims to blend cutting-edge AI technology with Warby Parker’s stylish eyewear designs. According to the announcement, Google has already invested $75 million to support product development and commercialization. An additional $75 million will be invested in exchange for equity if Warby Parker meets key performance milestones. The smart glasses will run on Android XR and feature multimodal AI, including capabilities powered by Google’s Gemini platform. The first line of products, which will include both prescription and non-prescription options, is expected to launch sometime after 2025. Google is clearly following a path similar to Meta, whose successful partnership with Ray-Ban manufacturer EssilorLuxottica helped launch a popular line of smart glasses. With Warby Parker’s strong brand presence and retail network, Google is positioning itself to bring smart eyewear to the mainstream market. This collaboration is part of a broader initiative by Google, which also includes partnerships with Samsung and Gentle Monster, to lead the future of AI-driven augmented reality. Source: (Techcrunch)
Google Secures 600 MW Solar Deal to Power Data Centers
Google has signed a major new clean energy deal, acquiring 600 megawatts (MW) of solar power to help power its growing network of data centers. The solar and battery storage projects will be developed in South Carolina by EnergyRe, a company currently managing around 16 gigawatts of energy projects across the U.S. This move is part of Google’s ongoing strategy to support its data infrastructure with clean energy, particularly as the demand for AI-driven computing continues to skyrocket. In recent years, the tech giant has significantly increased its renewable energy commitments. Last year, Google pledged $20 billion in investments with Intersect Power and TPG Rise Climate to develop zero-carbon power plants for its global data center operations. Earlier this year, Google also signed a separate deal for more than 700 MW of solar energy in Oklahoma with Leeward Renewable Energy. The push comes as Google aims to eliminate its entire carbon footprint from energy usage by 2030. While the company has made steady progress toward that goal, the exponential growth in compute demand—driven by AI and advanced data processing—has raised the stakes. In 2023 alone, Google contracted 4 gigawatts of clean energy. Google is not alone in this race. Microsoft and Meta have also ramped up their renewable energy acquisitions, with Microsoft purchasing 475 MW of solar in March 2025, and Meta securing 595 MW and 505 MW in back-to-back deals earlier this year. With data centers forming the backbone of modern AI applications, expect more major tech players to follow suit in scaling their clean energy footprints. Source: (Techcrunch)
NotebookLM App available for Android and iOS Ahead of I/O 2025
Just ahead of its highly anticipated I/O 2025 conference, Google has officially launched the mobile versions of NotebookLM, its AI-powered note-taking and research assistant. Previously available only on desktop since its 2023 debut, NotebookLM can now be accessed on Android and iOS devices, giving users greater flexibility to work and study on the go. NotebookLM is designed to help users understand and manage complex information more easily. The app offers intelligent features like smart summaries, the ability to ask questions about your documents, and even Audio Overviews — AI-generated podcasts based on materials you’ve uploaded. With offline support and background playback, users can stay productive even without an internet connection. The mobile app also includes the ability to: The launch comes just one day earlier than expected and is likely to be highlighted during Google’s I/O keynote. With this release, Google is making a strong move to position NotebookLM as a go-to tool for students, researchers, and anyone juggling dense information across multiple formats. Source: (Techcrunch)
Spotify iOS Users Can Now Buy Audiobooks In-App
In a major win for Spotify and audiobook fans alike, iPhone users in the U.S. can now purchase audiobooks and listening hour top-ups directly within the Spotify app. This marks a significant shift in Apple’s long-standing restrictions around in-app purchases and follows a recent U.S. court ruling that forced Apple to relax its grip on external transactions. A Smoother Audiobook Experience Until now, Spotify users had to jump through hoops — browsing the app for audiobooks but then switching to the web to make purchases. The update removes that friction. Now, users can: Each top-up adds 10 hours of audiobook time for $13, beyond the initial 15 hours that come with Spotify’s audiobook offering. This change is more than just a usability win — it’s a landmark moment in the ongoing battle over app store economics. Spotify had removed audiobook purchases from its iOS app back in 2022 due to Apple’s strict in-app payment rules. But the tide began to turn earlier this month when Spotify was approved to display subscription pricing and link users to external payment options. Spotify stated, “This change lowers the barriers for more users to embrace their first — or tenth — audiobook, while allowing publishers and authors to reach fans and access new audiences seamlessly.” With Apple slowly loosening its control over in-app commerce — under increasing legal and regulatory pressure — expect more developers to follow Spotify’s lead. For users, it means less frustration and more flexibility. For content creators, it’s a more open path to their audience. Source: (Techcrunch)
Apple Rebuilds Siri from the Ground Up Amid AI Struggles
Apple’s ambition to lead in generative AI has hit major roadblocks — and now, the company is betting big on a complete overhaul of Siri to catch up. According to a detailed report by Bloomberg’s Mark Gurman, Apple’s internal AI initiative, branded “Apple Intelligence,” has struggled with late starts, executive reluctance, and flawed product integration. What Went Wrong Gurman’s report reveals that Siri’s AI revamp — known internally as “LLM Siri” — was delayed by several missteps: The New Plan: Rebuild, Not Patch Apple is now completely rebuilding Siri around a large language model (LLM) architecture. The overhaul is being led by Apple’s AI team in Zurich, with the goal of creating an assistant that is: This shift also includes privacy-first AI training. Apple plans to use on-device differential privacy to generate synthetic data, comparing it against user patterns (like emails) without sending personal data to its servers. Giannandrea Steps Back Gurman reports that Giannandrea has been removed from Siri and AI product development roles. While he may remain at Apple, some insiders speculate he’s being moved “toward retirement” — though concerns remain about losing his research-heavy team. What’s Next? Apple’s new AI strategy may debut at WWDC 2025, with “LLM Siri” at the center. If successful, it could finally deliver on the long-standing promise of a smart, seamless digital assistant — one that Apple users have been waiting for far too long. Source: (TheVerge)
MIT Disavows AI Productivity Study Over Data Concerns
MIT has formally disavowed a widely discussed economics paper on artificial intelligence, citing serious concerns over the integrity and reliability of its findings. The paper, “Artificial Intelligence, Scientific Discovery, and Product Innovation,” authored by now-former doctoral student Aidan Toner-Rodgers, claimed that AI tools boosted material discoveries and patent output in a major science lab—while simultaneously reducing researchers’ job satisfaction. Initially praised by prominent MIT economists and Nobel laureates Daron Acemoglu and David Autor, the paper had gained significant attention across academic and tech circles. However, both economists have now issued a joint statement retracting their support, saying they have “no confidence in the provenance, reliability or validity of the data and in the veracity of the research.” Paper to Be Pulled from Journals The incident raises important questions about AI research ethics, academic oversight, and the influence of high-profile endorsements in rapidly evolving fields. It also underlines how even celebrated work can unravel under scrutiny—especially when claims intersect with commercial and scientific hype around AI. Source: (Techcrunch)
Heybike Alpha Review: A $1,699 Mid-Drive E-Bike that delivers big value
Heybike Alpha: The $1,699 All-Terrain E-Bike That Packs a Serious Punch If you’re in the market for an electric bike that balances power, comfort, and affordability, Heybike new Alpha might just be your ride-or-die. With a fat-tire, all-terrain build and a powerful mid-drive motor, the Alpha delivers a surprisingly smooth and natural ride for its $1,699 price tag. Whether you’re commuting through city streets or cruising over gravel trails, the Heybike Alpha offers premium performance at a mid-range price. Mid-Drive Motor, Maximum Value The Alpha is Heybike’s first model to feature its in-house Galaxy eDrive system, boasting a 500W Mivice mid-drive motor paired with a 680Wh battery. This setup delivers a fluid, pedal-assist riding experience that mimics a traditional bike, without the jerky acceleration often found in hub motor e-bikes. Pedal assist is responsive and powerful—thanks to 105 Nm of torque—and the addition of a Shimano 8-gear shifter allows you to fine-tune your power on different terrains, saving battery life in the process. Top speeds? 20 mph on throttle only, or up to 28 mph while pedaling. Built for Real-World Riding Range That Goes the Distance The UL-certified battery promises up to 60 miles per charge, and potentially more thanks to regenerative braking. One test ride covered 22 miles of hills without even dipping below 100% battery—a testament to the efficiency of Heybike’s system. Downsides to Consider The Heybike Alpha stands out as one of the smoothest mid-drive e-bikes in its price range. While there are a few build and usability quirks, its comfort, power, and value make it a compelling choice for anyone seeking a capable all-terrain ride. If you’re a bit handy—or don’t mind a trip to the local bike shop—this could be your new favorite way to get around. Source: (Techcrunch)
Spotify AI DJ Can Now Take Your Voice Requests
Spotify just took a big step forward in music personalization. The streaming giant has officially rolled out DJ Requests, a new voice-interactive upgrade to its AI-powered DJ feature—now live in over 60 countries for Premium users. This update lets users speak directly to their AI DJ, customizing music in real time without searching or scrolling. Want some electronic beats for your run? Need chill tracks for a rainy afternoon? Just say the word. How It Works: If you’re not sure what you’re in the mood for, just tap the DJ button again to skip to a new mix. While voice control in music apps isn’t entirely new, Spotify is positioning this as a way to turn listening from a passive experience into something much more interactive and personal. Think of it as having a music-savvy companion who just gets your vibe—no typing required. Though its long-term impact will depend on how well it handles nuanced commands, Spotify’s AI DJ Requests is a promising move toward truly dynamic music streaming. For now, it’s an exclusive treat for Premium users, but it could point to a future where your playlist literally listens to you. Source: (Mashbale)
