What Did Google Do This Week?
Google I/O was seismic for lots of reasons...
EDITOR NOTE: WDGDTW is moving to Beehiiv soon. Subscribe now before the price increases.
GOOGLE GOES AI EVERYTHING, EVERYWHERE, ALL AT ONCE
At I/O 2025, Google rolled out AI Mode in Search across the US, added Deep Search and agentic capabilities, and teased AI-powered shopping and try-ons. It also introduced Search Live, letting users talk to Search via camera or screen share, and integrated Gemini into Chrome with plans for full agentic browsing. Here’s the full 2-hour Keynote:
Google confirmed ads are coming to AI Mode, both within and below results, expanding AI Overviews to desktop. Google debuted NotebookLM Video Overviews, Firebase AI Logic tools, Project Mariner for web-browsing AI, and Jules, an asynchronous coding agent now in public beta.
In hardware, Google unveiled Gemini-powered XR glasses in collaboration with Samsung, Xreal, Warby Parker and Gentle Monster, and announced a $150M partnership with Warby Parker. Execs also demoed a prototype Android XR headset and backed Xreal’s Project Aura, the first Android XR glasses. The race is on with Meta to make these the must have/cheap enough gadget before OpenAI and Ives give people something new to lust after.
AI performance was a central theme: Gemini 2.5 Pro got Deep Think mode, while the experimental Gemini Diffusion model impressed with blazing token speeds. Google launched SynthID Detector to verify AI-generated content and confirmed it's tapping user data for Gemini personalisation.
New use cases also emerged: Google Meet’s AI translator mimics voice and tone, Gemini is coming to all Volvo models, and Android Auto will support video and browser apps. Google expanded its sovereign cloud offering in the EU, pitched Gemini as a tool for developers, and added Live Updates and shared payment options to its mobile ecosystem.
But risks are piling up. Google and Character.AI must face a wrongful death lawsuit, while the DOJ is investigating their partnership. A researcher exposed login credentials for Google and others, raising fresh security concerns. Not completely unexpected for a company of Google’s size and breadth, but still, not a look Google wants.
Finally, AI content quality drew scrutiny. Veo 3’s video generation was lauded for realism, but raised alarms about misinformation and manipulated dialogue.
SO WHAT?
So did Google I/O 2025 position Google as an AI-native platform company, not just a search engine with add-ons? Kinda. Google is no longer just demoing AI, it’s showing us how the company is operationalising it across every product. Gemini isn’t a feature but a framework, spanning infrastructure, user interfaces and monetisation. The company is pushing to redefine itself as an AI-native platform, not just a search engine with bolt-ons. But this centralisation creates new exposure. Personalised agents require personal data. Real-time answers blur editorial lines. XR glasses promise ambient access, but also ambient risk. At the same time, regulatory and reputational hazards are scaling in lockstep. Google’s ambition is clear: to own the next interface paradigm and the AI developer stack. The question now is whether Google can do that faster than the trust, and compliance, gap widens.
Stay ahead with the full 'What Did ____ Do This Week?' ecosystem.
Choose from OpenAI, Amazon and the new Google version.