In a significant development for digital technology and user privacy, Google has announced a new data policy for its artificial intelligence chatbot, Gemini. Effective September 2, 2025, Google will begin utilizing a sample of user chats and uploaded content—including files, photos, videos, and even screenshots—to train and enhance its AI models. This move, while intended to boost Gemini’s accuracy and usefulness, raises substantial questions about data privacy, compelling users to take action to protect their personal information.
The Core of the New Policy
The upcoming change means that interactions with Gemini, previously thought by some to be private, will now actively contribute to the development of Google’s artificial intelligence. The company confirmed via email that this feature will be enabled by default for all users aged 18 and over. This places the onus squarely on the individual to adjust their settings if they wish to prevent their data from being used for AI training. Google clarifies that if a user had previously disabled the ‘Gemini Apps Activity’ setting, the new ‘Keep Activity’ setting, which is the rebranded version, will remain off, respecting their prior choice.
AI models, particularly large language models (LLMs) like Gemini, require vast amounts of diverse data to learn language patterns, reasoning, and context. Publicly available datasets, while helpful, are often insufficient to capture the nuances of real-world queries and human communication. By incorporating a sample of user interactions, Google aims to refine Gemini’s ability to understand prompts and deliver more relevant and helpful responses, driving forward advancements in conversational AI.
Understanding the Implications for User Privacy
While Google asserts that conversations sent for service improvement are disconnected from individual accounts before being shared with its providers, the prospect of personal interactions being analyzed for AI training has sparked considerable privacy concerns. Users frequently share deeply personal details with chatbots, ranging from health inquiries to financial advice or emotional struggles. The idea that such sensitive information, even if anonymized, could be stored and analyzed for model improvement is unsettling for many.
This policy update also highlights the broader balance that tech companies must strike between advancing AI capabilities and safeguarding user privacy. Google’s longstanding commitment to a “privacy-by-design” approach and its responsible AI principles emphasize integrating strong privacy safeguards, data minimization techniques, and providing transparency. However, the default-on nature of this new setting for user uploads presents a direct challenge to the expectation of privacy by default.
How to Reclaim Your Data Before September 2
Fortunately, Google has provided clear mechanisms for users to opt out of this data usage. Given the September 2 deadline, prompt action is advisable for those who wish to maintain stricter control over their Gemini data.
For Desktop Users:
1. Visit `gemini.google.com` and sign in to your Google account.
2. Click on the three-bar menu icon on the left-hand side to open ‘Settings & help’.
3. Select ‘Activity’. This will take you to a new settings page.
4. Locate the toggle for ‘Gemini activity’ (which will soon be renamed ‘Keep Activity’) and switch it off.
For Mobile Users (Gemini App):
1. Open the Gemini application on your smartphone.
2. Tap on your profile picture, typically located in the top-right corner of the screen.
3. Navigate to ‘Gemini apps activity’ (soon to be ‘Keep Activity’).
4. Toggle the setting off.
Managing Past Activity and Data Retention:
Users also have granular control over past data. You can delete specific activity items or entire conversation histories from your Gemini Apps Activity page (`myactivity.google.com/product/gemini`). Options typically include deleting activity from the last hour, last day, all time, or a custom date range. While disabling the feature prevents future conversations from being saved for training, Google notes that, for service and feedback purposes, data may be temporarily retained for up to 72 hours before permanent deletion, even if activity saving is turned off.
Furthermore, Gemini Apps Activity is set to auto-delete anything older than 18 months by default. Users can adjust this auto-delete period to 3 months, 36 months, or choose to turn off auto-delete entirely for indefinite retention. For users managing multiple Google accounts, it’s crucial to repeat the opt-out process for each individual account to ensure comprehensive privacy.
The Evolving Landscape of AI and Privacy
This latest news regarding Gemini’s data policy underscores the ongoing tension between technological advancement and individual data sovereignty. As AI systems become more integrated into daily life, these policy updates will likely become more common. Google’s approach, while offering an opt-out, necessitates active user engagement to safeguard personal data in the evolving world of AI. Users are encouraged to regularly review their privacy settings across all digital platforms, especially as AI capabilities continue to expand.
