Google shut down Project Mariner and folded its tech into Gemini
Google quietly killed Project Mariner on May 4, 2026, absorbing the experimental browser-automation tool into its Gemini AI platform rather than letting it die outright. Launched in December 2024, Mariner could handle up to 10 simultaneous web tasks — booking hotels, comparing flight prices, filling out forms — all without the user lifting a finger. Its closure marks another chapter in Google's long history of sunsetting experiments, though this time the technology didn't simply vanish.
What Mariner was
Project Mariner was a browser agent: software that could navigate websites autonomously, interact with buttons and filters, and complete multi-step tasks on behalf of the user. It was an ambitious idea, but it lived in Google Labs as an opt-in experiment rather than reaching mainstream products. Wired (Maxwell Zeff) first reported the shutdown, noting the project's official landing page confirmed it had ceased operating as a standalone product.
NEW: Google quietly shut down Project Mariner yesterday, the web-browsing AI agent it highlighted onstage last year at Google IO.
— Max Zeff (@ZeffMax) May 6, 2026
I reported for WIRED, nearly 2 months ago, that Google had moved staffers off the Project Mariner team as it responded to OpenClaw-style agents. pic.twitter.com/WMBago74vr
Google says Mariner's core algorithms have been integrated into Gemini Agent — the AI assistant's task-automation layer — and into Google AI Mode in Search. A related feature, Auto Browse, rolled out in Chrome in early 2026, letting the browser navigate complex web flows without human input. The architectural similarity to Mariner is hard to miss, even if Google hasn't drawn a direct line between the two.
The US-only catch
Auto Browse is currently available only to AI Pro and Ultra subscribers in the United States. There is no announced timeline for a UK or EU rollout, and questions around browser agents auto-filling forms or handling authentication under GDPR remain publicly unaddressed by Google.
That matters because The Verge has noted the wider browser-agent market is already competitive. OpenAI shipped ChatGPT Atlas in October 2025 and Perplexity launched Comet in July 2025 — both offering agentic browsing to users outside the US. Google's decision to fold Mariner into existing products rather than ship a standalone competitor suggests it chose platform consolidation over disruption.
What this means going forward
The closure fits a broader industry shift. API-centric agents — tools that work through structured interfaces rather than visually "reading" a webpage — are increasingly preferred for their reliability and lower compute cost. Mariner's visual browsing approach, while impressive in demos, faced real-world friction. Integrating the technology into Gemini lets Google quietly retire the hard parts while keeping the narrative positive: the project didn't fail, it "voyaged to other products."
For most users, the practical upside arrives when Auto Browse expands beyond the US — at which point features like automated booking or form completion could become a standard part of using Chrome and Gemini together.
