Google is pulling the plug on one of its experimental AI projects.
Google has shut down Project Mariner, an AI browser assistant designed to browse websites, click through pages, and complete tasks for users automatically.
This was not a normal chatbot.
The AI could interact with websites almost like a human using a browser. It could research information, move through pages, and handle multi step tasks online without users doing everything manually.
That made it part of the growing push toward “agentic AI,” systems that do things instead of only answering questions.
At one point, Google expanded the project to handle multiple tasks at once, showing how serious the company was about browser automation.
But now the standalone project is gone.
According to reports, Google is moving some of Mariner’s technology into other products instead of keeping it separate. That means parts of it may still appear inside Gemini and future Chrome features.
The shutdown also highlights a bigger problem with AI browser agents.
Giving AI control over real websites sounds powerful, but it also creates security and reliability risks. If the AI misunderstands instructions or interacts with the wrong page, the consequences become much more serious than a simple chatbot mistake.
Still, Google clearly believes browser AI is still important.
The company is just changing how it delivers it.
And that says a lot about where AI tools are heading. Companies no longer want AI that only talks. They want AI that can actively perform tasks for users across the internet.
So the real question is not whether browser AI assistants are coming.
It is whether people will trust them enough to let them operate online on their behalf.
