Google’s Gmail Now Reads Emails and Attachments to Train AI by Default, Unless Users Opt Out
In a significant and somewhat controversial update, Google has started automatically opting Gmail users into a program that allows the company to scan private emails and attachments to train its artificial intelligence (AI) models, including its Gemini AI system. This new policy change means, unless users actively opt out, their private communications could be used to enhance Google’s AI-powered features such as Smart Compose, Smart Reply, and predictive text functionality.
The rollout of this update has been described as quiet and gradual, catching many users off guard as these settings appear to be enabled by default in some accounts. These AI training procedures analyze the content of emails and attachments not just for improving responsiveness but also to deepen Gemini’s contextual understanding of user communications across Gmail, Chat, Meet, and other Google Workspace services.
Privacy Concerns and Opt-Out Process
While Google assures users that the data is anonymized and handled securely during training, privacy advocates and some users express concern about the implications of sharing sensitive or confidential emails without explicit, informed consent. The fact that users must actively disable these AI training features to safeguard their data has sparked backlash, emphasizing the need for greater transparency and control.
To opt out completely, users must adjust settings in two different parts of their Google account. The reason for this dual setting is because Google segregates “Workspace smart features” (which cover email, chat, and meet interactions) from general smart features used across other Google apps. Both toggles need to be disabled — a process that some find non-intuitive and insufficiently highlighted by Google.
Steps to Disable AI Training on Gmail
- Open Gmail settings by clicking the gear icon.
- Navigate to the “Smart features” section and turn off the options that allow AI-powered assistance.
- Further, within Google Workspace settings, locate and disable “Workspace smart features” to ensure no data from workspace apps is used for AI training.
Experts recommend verifying these settings regularly as Google continues to roll out AI enhancements and refinements to its platforms.
Legal and Industry Context
This update coincides with increasing scrutiny over big tech’s use of personal data for AI development. Reports indicate that over 1.8 billion Gmail users worldwide may have been automatically enrolled in these AI training programs, with minimal notification. The issue has already triggered class-action lawsuits in some jurisdictions, including California, over potential privacy violations.
From an industry perspective, Google maintains a commitment not to use customer data for training generative AI models outside of customer permission or instruction, particularly within Google Workspace data environments. However, the recent policy shift puts individual Gmail users at more immediate risk unless they manually intervene.
Broader Implications for Gmail Users
The widespread use of AI technologies like Gemini in Gmail aims to provide a smarter, more personalized user experience — such as enhanced writing suggestions or contextual email management. Nonetheless, the tradeoff involves a delicate balance between convenience and privacy, with many users uneasy about automated processes analyzing their private correspondence.
Additionally, internal data retention practices mean some anonymized interactions may be kept for up to three years for quality review, raising further questions on long-term privacy and data governance.
For users concerned about personal data security, experts advise immediate review and adjustment of Gmail AI settings to maintain control over how email data is used.