Realtime AI News
Google opens early access to a Google Home MCP server so AI agents can control your home
Google is launching early access to a new MCP server for Google Home that lets AI agents such as Claude and ChatGPT control connected devices in natural language. The server also exposes camera summaries and smart home activity, turning the assistant people already use into the control layer for the home.

Smart home control is moving out of first-party apps and into whichever AI assistant a person already talks to. On September 16, TechCrunch reported that Google is launching early access to a new MCP server for Google Home, allowing AI agents such as Claude and ChatGPT to control connected devices.
The server does more than flip switches. According to the report, agents can also review camera summaries and access smart home activity, and the interaction happens in natural language rather than through device-by-device menus inside an app.
MCP, the Model Context Protocol, is a standard way for AI assistants to connect to external tools and data sources. By shipping an MCP server for Google Home, Google effectively turns a household's devices and activity history into tools that outside assistants can call, instead of keeping that capability inside its own assistant.
The real signal is where the competition is heading. For years people controlled devices through separate apps or a vendor's voice assistant; once Claude or ChatGPT can drive those same devices, the front door becomes the assistant someone uses every day, and the hardware platform's value shifts toward device coverage, data, and the ability to actually execute actions.
For Google, this is a comparatively open posture: instead of insisting that users stay inside its own assistant, it is letting third-party assistants reach into the home. That helps Google Home keep its position as infrastructure in the contest over which assistant becomes the brain of the home.
It is worth noting that this is early access, not general availability, and the report does not give a timeline for a broader rollout. Early access phases are typically used to test permission models, reliability, and safety boundaries, especially when an outside assistant can read camera summaries.
Several things are worth watching next. Which assistants get access first, and how many device categories they cover; how permissions, logging, and audit trails are handled for privacy-sensitive capabilities such as camera summaries and activity history; and whether other smart home platforms follow with their own MCP servers, making this the standard way agents get into the home.
Sources
Why it matters
Putting Google Home behind an MCP server moves the smart home fight from apps and hardware toward assistant ecosystems. If the approach spreads, permission and privacy design, not device count, becomes the test of whether agent-controlled homes feel trustworthy.
Nearby Updates
All09/17, 01:05
TypeSafe launches Jev, a non-chat AI model it claims is 193x faster than Claude
TypeSafe has launched an AI model called Jev, positioned as a non-conversational system that the company claims runs 193 times faster than Anthropic's Claude. The claim reframes the model's value around speed rather than chat, though the basis for the 193x comparison has not been disclosed.
09/17, 00:30
Anthropic merges Claude chat and Cowork into one interface
Anthropic is folding Claude chat and Cowork into a single front end so users no longer have to choose a tab, with requests routed to the right capability automatically. The release also adds presentation and document features, reaching Pro and Max subscribers first before the free and team tiers.
09/17, 00:27
Spain logs its first data breach blamed on an autonomous AI agent
Spain's data protection agency has received what it calls the country's first report of a personal data breach carried out by an AI agent, which used a well-known large language model and exploited an application flaw to change personal data and view invoices. The AEPD said the details come solely from the affected organisation and still need to be analysed.
09/17, 00:02
Microsoft AI chief warns Anthropic's Claude training carries disaster risk
Microsoft's top AI executive has publicly warned that the way Anthropic trains its Claude models could carry catastrophic risk, according to a report from Unite.AI. The statement turns a long-running internal safety debate into an open, competitive public argument between two major AI players.