Key Takeaways
Google firmly denied using Gmail content (emails and attachments) to train its AI models, including Gemini AI.
Viral claims that Google changed privacy policies for AI development purposes are inaccurate, according to the company.
Share this article
Google denied claims that it uses Gmail data to train its AI models, following viral reports suggesting the tech giant had changed its privacy policies to access user emails and attachments for AI development purposes.
The company clarified that Gmail’s existing smart features, such as spell checking and smart replies, use data for personalization but remain separate from AI model training. “We do not use Gmail content to train our Gemini AI,” Google stated, emphasizing that no policy changes have occurred regarding AI training data.
Recent discussions emerged after some users reported being unexpectedly re-enrolled in Gmail’s smart features, which analyze user data to provide conveniences like attachment scanning and automated responses.
Gmail’s smart features have long utilized user data for personalization purposes, but Google maintains these functions operate independently from its AI model development. The company highlighted that users retain privacy controls over data usage through their account settings.
The clarification comes as privacy concerns grow around major tech companies’ data collection practices for AI development, with users increasingly scrutinizing opt-out options for various data usage features.
