Two very different strategies for AI video tools played out this week. Google quietly rolled AI-generated video and music into Google Vids - the workspace tool most people didn't know existed. Meanwhile, OpenAI shut down public access to Sora, the video model everyone was talking about six months ago.
The contrast is sharp. One company is putting AI video generation directly into the hands of office workers making quarterly review presentations. The other is pulling back from public release entirely.
Google Vids gets the AI treatment
Google Vids sits inside Workspace alongside Docs, Sheets, and Slides. It's designed for quick video creation - the kind of thing you'd use to explain a process to your team or recap a project milestone. Nothing fancy. Utilitarian.
The new AI features let you generate video clips from text prompts and create background music to match your content. The idea is simple: describe what you need, get a usable video, drop it into your presentation. No film degree required.
It's not cinematic. It's not designed to be. Google is betting that most video content created at work doesn't need to be beautiful - it just needs to exist and communicate clearly. A training video. A product demo. An update for stakeholders who won't attend the meeting.
The strategic move here isn't the technology itself. It's the delivery mechanism. Google isn't launching a standalone AI video product that people need to discover, sign up for, and learn. They're sliding it into a tool that's already part of the daily workflow for millions of Workspace users. You don't go looking for AI video generation. You're making a presentation and realise it's just... there.
OpenAI pulls Sora from public access
Sora launched with genuine wow factor. The demo videos were strikingly good - far better than anything else publicly available at the time. OpenAI positioned it as a research preview, available to a limited number of users.
Now it's gone. Public access has been withdrawn. No timeline for when it might return.
The official reason focuses on safety and responsible deployment. That's standard language for OpenAI, and it's not wrong - video generation at scale raises legitimate concerns about deepfakes, misinformation, and content authenticity. But there's also a practical angle nobody's saying out loud: compute costs for video generation are brutal. Running Sora at any real scale would be expensive in ways that text generation isn't.
The pullback suggests OpenAI is rethinking how - or whether - to make Sora available publicly. They've done this before with other models, holding back features or access when the economics or safety story doesn't line up with user expectations.
Two philosophies on distribution
What's interesting here isn't just the technology. It's the approach to getting it into people's hands.
Google is embedding AI tools into existing workflows. You're not learning a new platform or adding another subscription. You're using Google Docs, and suddenly there's an AI assistant built in. You're making a presentation in Vids, and video generation is just one more feature alongside transitions and text boxes. The friction to adoption is near zero.
OpenAI has historically taken the opposite approach: build the best possible model, release it as a standalone product, let developers and creators figure out what to do with it. ChatGPT worked because the use case was obvious and the interface was dead simple. But video generation is different. The use cases are more specific. The expectations are higher. And the costs of running it are steeper.
The result is that OpenAI has the more impressive technology on paper, but Google has the distribution advantage. A mediocre tool that's embedded in your daily workflow beats an excellent tool that lives behind a waiting list and a separate login.
For business owners and teams, the practical implication is straightforward: if you're already using Google Workspace, you now have access to AI video tools without changing anything about how you work. If you were waiting for Sora access to create marketing content or training videos, you're back to looking for alternatives - or learning to use the tools that are actually available.
The real question is whether OpenAI plans to bring Sora back as a premium product integrated into ChatGPT, or if this is a longer-term retreat while they figure out the economics. Either way, Google just took a lead by showing up where people already are.