AI News: These Google Updates Are Dividing People
Matt Wolfe showed 21 AI tools in this video on . 5 are verified available in the EU. Each row below jumps to the exact second. Part of week 2026-21.
Tools shown in this video, with timestamps
Smaller, faster model in the Gemini 3.5 family; roughly on par with Gemini 3.1 Pro and Opus 4.7 in intelligence but over twice as fast, priced at $1.50 input / $9 output per million tokens. Gemini 3.5 Pro is coming later.
Any-input-to-video model that edits video with natural language, keeps character consistency, supports avatars, and is grounded in world knowledge. Creator tested generation, editing and avatar features with mixed results; 10-second clips, paid plans only.
Live demoGoogle's always-on agent running on Google Cloud with access to Gmail, Calendar and Drive, supporting recurring tasks, MCP connectors and importable skills. Rolling out to trusted testers, then Ultra subscribers.
Search becomes a longer prompt box defaulting to AI mode, adds search agents that alert on new info, and writes interactive code visuals in responses. Creator discusses concerns about reduced traffic to publishers; also mentions Ask YouTube.
Live demoA shopping cart that works across merchants, flags lowest prices and incompatible items, and lets agents check out via a new agent payments protocol.
New version of Google's agentic IDE, whose look closely resembles OpenAI's Codex app, whereas the old version was a VS Code fork.
Google Labs brand-building tool that added brand books and the ability to build a complete website from a brand kit.
Vibe design tool that builds a website from a prompt in real time.
Suite of scientific tools and experiments, including science skills integrating over 30 life science databases such as UniProt, AlphaFold, AlphaGenome and InterPro.
Interactive world generation now grounded in real locations via Google Street View. Creator demoed a monkey on roller skates at Petco Park and called it impressive.
Live demoAI glasses with Gentle Monster and Warby Parker in audio/camera and display versions. Audio versions launch in fall; they handle messages, calls, navigation, photos and translation.
Watermark standard for AI content now being adopted by OpenAI, ElevenLabs, Kakao and others, letting Gemini detect AI-generated images, audio and video.
ChatGPT feature connecting bank accounts via Plaid to give finance guidance grounded in your data. Creator voices privacy and advertising concerns.
Spotify and Universal Music Group licensing deal for fan-made AI covers and remixes with artists sharing revenue; requires Premium.
Lets users generate and schedule private AI-generated audio episodes from a prompt, plus ask questions about podcasts in real time.
Alexa+ can generate podcast-style episodes on demand on any topic in minutes.
Cursor's coding model, near Opus 4.7 and GPT 5.5 on benchmarks at under a dollar per task.
Model family generating up to 6-minute songs and sound effects; small and medium models are open weights while the largest 2.7B model is closed.
Krea added LoRA training for Krea 2 so generated images can match a specific style or character.
New briefs feature researches meeting participants, such as via LinkedIn, before a call to prepare you. Creator uses Granola for meeting notes.
Boston Dynamics showed a robot picking up and carrying a whole fridge; the fridge showed some damage.
- Here's the AI News you might have missed this week. If your team lives in Slack, check out @getviktor_com for free atref.viktor.com
- Explore AI Tools & Newsfuturetools.io
- 100 I/O Announcementsblog.google
- Gemini 3.5blog.google
- Gemini Omniblog.google
- Agentic Gemini Appblog.google
- Search I/O Updatesblog.google
- Antigravity 2.0antigravity.google
- Pomelli AI Capabilitiesblog.google
- Stitch Real Time Designblog.google
- Gemini For Scienceblog.google
- Project Genie Expansionblog.google