Google Adds Memory Feature to Gemini Chatbot for Premium Web Users
Google's Gemini chatbot now allows premium web users to save personal preferences and context for more tailored conversations. The company promises these memories remain private and are not used to train AI models.
Google introduces a new memory feature to its Gemini chatbot, allowing the AI to remember details about a user's life, work, and personal preferences. This tool enables the chatbot to provide highly tailored responses, such as recommending restaurants based on previously saved dietary preferences. Users can prime the system with specific instructions, like avoiding jargon or specifying programming languages, to customize their ongoing interactions.
The memory capability is currently rolling out exclusively to subscribers of the $20-per-month Google One AI Premium plan. For now, the feature remains limited to the web client and supports only English-language prompts, with no immediate availability on the Gemini mobile apps for iOS and Android. Users retain full control and can toggle the feature off or manually delete their saved memories at any time.
Google emphasizes strict privacy measures, confirming that saved memories are never shared or used to train the underlying AI model. However, the introduction of such features carries known security risks, as similar memory functions in rival chatbots have previously been exploited by hackers to plant false memories and steal user data. As AI assistants become more personalized, maintaining robust guardrails remains a critical challenge for developers.