Google denies claims of using Gmail data to train AI models

Key Takeaways

Google firmly denied using Gmail content material (emails and attachments) to train its AI models, together with Gemini AI.
Viral claims that Google modified privateness insurance policies for AI growth functions are inaccurate, in accordance to the corporate.

Share this text

Google denied claims that it makes use of Gmail data to train its AI models, following viral experiences suggesting the tech big had modified its privateness insurance policies to entry person emails and attachments for AI growth functions.

The firm clarified that Gmail’s present sensible options, similar to spell checking and sensible replies, use data for personalization however stay separate from AI mannequin coaching. “We do not use Gmail content to train our Gemini AI,” Google said, emphasizing that no coverage adjustments have occurred concerning AI coaching data.

Recent discussions emerged after some customers reported being unexpectedly re-enrolled in Gmail’s sensible options, which analyze person data to present conveniences like attachment scanning and automatic responses.

Gmail’s sensible options have lengthy utilized person data for personalization functions, however Google maintains these features function independently from its AI mannequin growth. The firm highlighted that customers retain privateness controls over data utilization via their account settings.

The clarification comes as privateness considerations develop round main tech firms’ data assortment practices for AI growth, with customers more and more scrutinizing opt-out choices for numerous data utilization options.