Google has introduced a new feature that automatically saves every piece of media uploaded during searches to train its artificial intelligence models. Images uploaded for reverse image searches, audio from Google Translate conversations, and similar media are now stored in the Search Services History, a section of the account that many users found already active upon first visit. This change, announced via email on June 23, offers users more control over saved data, but raises privacy concerns and places the burden on users to opt out.
An extensive and automatic data collection
The new setting significantly expands the amount of data Google retains from interactions with its services. It is no longer just text inputs but includes images, files, audio, and video recordings. These include photos taken with Google Lens, voice searches, content uploaded during Search Live, and pronunciation exercises in Google Translate. Google states that this data helps deliver more relevant results and improves visual and voice search quality. However, Google's own documentation specifies that saved media may be used for AI model training, with a retention period of up to four years even after the original activity is deleted.
Sponsored Protocol
How to disable the Search Services History
To turn off this feature, users must visit the My Activity page and select the Search Services History tab. Here, they can disable the entire setting and delete accumulated activity. The crucial step is to uncheck the Save media box; otherwise, uploads will continue to be stored for AI training. Google itself warns that once media is used for training, it is disconnected from the account and retained for up to four years, even if the original activity is deleted. Therefore, it is advisable to act immediately to prevent personal data from entering the machine learning pipeline.
Sponsored Protocol
Privacy debate and the burden of opting out
Google's decision to enable this feature by default has drawn criticism from privacy experts. Thorin Klosowski, senior security and privacy activist at the Electronic Frontier Foundation, notes that Google is in a unique position due to the wide range of services people use daily, creating inertia that makes switching to alternatives difficult. Ben Winters, director of AI and privacy at the Consumer Federation of America, highlights that this change imposes an extra calculation on consumers, who must decide if they feel comfortable using a tool they have relied on for years. While companies like OpenAI upgrade their models to reduce hallucinations, Google pursues more massive data collection. Although fears about AI's impact on jobs persist, SignalFire data shows an increase in engineers, but the privacy issue remains central.
Sponsored Protocol
A four-year retention window for training data
One of the most concerning aspects is the retention duration for training data. Even if a user decides to delete their activity, media already used for training remains in the system for up to four years. This means a single image uploaded for a reverse search could influence Google's AI models for years, with no possibility of removal. For more details on how Google Search works, see the Wikipedia page on Google Search.
Ultimately, Google's new feature represents a further step toward extensive use of user data for AI training. Opting out is possible but requires a conscious action by the user, who must navigate account settings. As experts point out, the responsibility should not always fall on the consumer; companies should adopt transparent opt-in models.
Source: https://www.wired.com/story/how-to-opt-out-of-google-search-new-ai-data-training
