Conversations with Gemini archived for years by Google by default

Google, the tech giant, has disclosed that it saves users’ conversations with Gemini, its GenAI family of apps, for a default period of three years. This means that all conversations with the Gemini chatbot, which include data such as languages used, devices used and user location, are archived and stored for a long period of time.

How Gemini works

Gemini is a chatbot developed by Google that usesartificial intelligence to provide aninteractive conversational experience for users. It can be used on a variety of platforms, including the Web, Android and iOS. However, what might surprise many users is the fact that Google saves all conversations with Gemini for a period of three years.

According to a Google support document, human annotators regularly read, tag and process conversations with Gemini, even if these conversations are “disconnected” from users’ Google accounts. It is unclear whether these annotators are internal or external to the company, but this could be relevant with regard to data security. Google retains these conversations along with related data to improve the service provided by Gemini.

Control over saved information

Google offers users some controls over the information saved by Gemini. You can turn off Gemini’s app activity in the “My Activity” panel of Google. This option prevents future conversations with Gemini from being saved in the user’s Google account for review. However, even when Gemini app activity is turned off, conversations are still saved in the Google account for up to 72 hours to ensure the security and reliability of Gemini apps.

In addition, users also have the option to delete individual prompts and conversations from the Gemini app activity screen. This provides a degree of control over conversation archives and allows users to protect the privacy of their information.

The privacy challenge of using GenAI data.

Google’s data retention policy for Gemini does not differ much from that of other GenAI players. For example, OpenAI saves all ChatGPT conversations for 30 days, regardless of whether the ChatGPT conversation history feature is enabled or not. However, liberal data retention policies such as these have attracted the attention of regulators in the past.

The Federal Trade Commission (FTC) has requested detailed information from OpenAI on how the company selects and protects data used to train its GenAI models, including consumer data. In addition, the Italian Data Protection Authority said that OpenAI has no “legal basis” for the massive collection and storage of personal data to train its GenAI models.

These examples demonstrate that companies developing GenAI tools face the challenge of balancing user privacy with developing models that rely on user data to continuously improve.

Growing privacy concerns

As the use of GenAI tools increases, organizations are increasingly concerned about privacy risks. A recent survey conducted by Cisco revealed that 63 percent of companies have established restrictions on what data can be entered into GenAI tools, while 27 percent have banned the use of such tools altogether. In addition, 45 percent of employees have entered “problematic” data into GenAI tools, including employee information and non-public files about their employer.

Although some companies such as OpenAI, Microsoft, Amazon and Google offer GenAI products for businesses that do not retain data for any period of time, consumers often do not enjoy the same levels of privacy protection.