Google Gemini Memory Feature Poses Data Theft Risk
Google's Gemini AI chatbot introduced a memory feature that stores user information persistently. Cybersecurity researchers warned the feature could allow hackers to steal data or inject false memories, though no actual attacks have been reported.
What happened
Google's AI chatbot Gemini introduced a new memory feature that allows the system to retain and recall information about users' habits and personal lives.
The feature is available only to paid subscribers for $20 per month and currently supports English-language queries only.
Although users can disable the memory feature at any time, stored data remains in the system until manually deleted by the user.
Reported impact
Cybersecurity experts warned that skilled hackers could potentially exploit Gemini's memory function to steal user data or inject false memories into the system.
Index analysis
The persistent storage of user data even after the feature is disabled creates a potential vulnerability for unauthorized access.
Classification
Sources and evidence
Sources not yet published
This record was researched and written by the Index from original reporting. Its sources are still being verified by an editor, and they are listed here once that check is complete. Until then the record carries no reviewer.