Google adds automatic memory and temporary chat controls to Gemini
Google has begun rolling out automatic memory in Gemini AI, allowing the assistant to remember details from past interactions by default. The update also introduces a “Temporary Chat” mode that does not store or use conversations for training and expires after 72 hours. The changes aim to balance personalization with stronger privacy controls.
Google is enabling automatic memory for Gemini 2.5 Pro in select markets, with availability expanding to Gemini 2.5 Flash. The feature stores information such as names, preferences and recurring topics to tailor responses across sessions. Users can view and manage this data under a new “Personal Context” section in settings, including the option to disable memory entirely.
The company is also launching “Temporary Chats” for one-off or sensitive conversations. These sessions remain isolated from the memory system, are excluded from model training and automatically expire after three days. This option is available alongside existing chat history controls.
In a related change, “Gemini Apps Activity” will be renamed “Keep Activity” starting 2 September. This update will also allow Google to sample file and photo uploads for quality improvements, with options for users to manage or delete stored activity.
Google says the rollout will prioritize transparency and user control, with prompts informing users when information is being stored or when a chat is temporary. The combination of personalization and enhanced privacy tools positions Gemini to better compete with AI assistants that already offer similar capabilities.
Pure Neo Signal:
Comment
This week, Anthropic’s Claude and Google’s Gemini finally delivered memory features that track past conversations. ChatGPT introduced a similar ability only a few weeks earlier. It is progress, but what matters is what you can actually do with that memory.
OpenAI’s GPT-5 launch delivered horsepower, long context windows, routed reasoning, safety improvements, and versatile coding. It is not showing off much in terms of real-world workflows. In contrast, Anthropic and Google are wrapping practical tools around their models. Claude does not just remember. It is backed by Claude Code, an agentic assistant that lives in your terminal, understands your whole codebase, manages multi-file edits, runs tests, and acts autonomously. Google’s NotebookLM weaves AI into the research process, letting you ask questions of your documents, summarize dense material, and create AI-generated podcasts or featured learning notebooks.
Here is the rub. GPT-5 shows promise and power. But Claude and Gemini offer tangible tools you can use today, whether you are coding, researching, or organizing information.
We are entering an era where utility trumps hype. OpenAI may have the brain. Anthropic and Google are building the ecosystems. And right now, ecosystems get work done.
We love
and you too
If you like what we do, please share it on your social media and feel free to buy us a coffee.