Google has once again expanded the scope of data it uses to train its artificial intelligence models, this time automatically opting users into sharing images, videos, and voice searches. Unless you manually disable the feature, any media you upload or generate while using Google's search services may be fed into the company's large language models (LLMs). Here's what changed, why it matters, and how to opt out.
What Google Changed
In a quiet policy update emailed to users last month, Google revised its Search Services History documentation to include media files as training material. The updated text states: "You can choose whether Google saves media to your Search Services History when you're signed in. Saved media includes your images, files, audio, and video from your interactions with Search services. Your media may be used to improve your experience on Google services, like letting you revisit your past visual searches. Saved media may be used to develop and improve Google's AI models and technologies, as well as the Google services that use them."
Previously, Google's AI training primarily relied on text-based search queries, web content, and anonymized usage patterns. Now, the company is leveraging richer, more personal data: photographs taken with your phone, screenshots from feature like Circle to Search, voice commands entered via Search Live, and even recordings made while using Google Translate for pronunciation practice. This shift marks a significant expansion of the data Google considers fair game for AI development.
Why This Raises Privacy Concerns
The immediate consequence is a reduction in user privacy. When you upload a photo of your dog to identify its breed, or use voice search to ask about local restaurants, that media is no longer just a temporary request—it becomes part of a permanent dataset used to train algorithms. Over time, these datasets can reveal intimate details about your life: your appearance, your surroundings, your voice patterns, and your conversations.
Consider a scenario where you use a Google chatbot to discuss a sensitive topic, such as dissatisfaction with your current job. The chatbot records your words and images you might share. That data is then used to refine the LLM. Later, your employer asks the same AI a question about employee morale. Because the model was trained on your private conversation, the employer's query might surface similar language or concepts, potentially exposing your identity or your complaints. While Google claims it anonymizes data, the line between anonymous and identifiable becomes blurred when personal media is involved.
Voice data is especially concerning. Your voice is unique biometric information. If a malicious actor gains access to the training dataset or the model itself, they could synthesize audio clips that sound exactly like you saying things you never said. This opens the door to impersonation fraud and blackmail.
Historical Context: Google's Data Practices
This is not the first time Google has faced criticism over data collection. In 2021, the company was fined $100 million by French regulators for placing tracking cookies without explicit consent. In 2023, it settled a $391 million privacy lawsuit with 40 U.S. states over location tracking practices. The current change continues a pattern of expanding data use under the guise of improving services, while often making opt-out options less visible.
Google's AI ambitions are clear. The company is investing heavily in its Gemini model and competing with OpenAI, Meta, and Microsoft. Training sophisticated LLMs requires vast amounts of diverse data. Media files—photos, videos, audio—provide richer context than text alone. However, critics argue that the company should obtain active, informed consent before using such personal data, rather than relying on a default opt-in model buried in policy updates.
How to Opt Out (Step-by-Step)
Fortunately, you can disable the feature through two key settings: Search Services History and Search Services Personalization. Below are detailed instructions.
Disable Search Services History
- Open the Google app on your smartphone (Android or iOS).
- Tap your profile picture in the top-right corner.
- Select "Search history" from the menu.
- You'll see an option labeled "Save media." Make sure the toggle is switched off (gray).
- Below that, tap on the "Search Services History" entry itself. If it shows as "On," tap it to turn it off. Confirm the change when prompted.
Disable Search Services Personalization
- Return to the Google app's main menu (tap your profile picture again).
- Choose "Search personalization."
- Tap the top entry on that page.
- From the resulting screen, toggle the service off. This limits Google's ability to tailor results based on your activity, but also prevents media from being used for training.
Alternative: Use the My Activity Page
If the in-app options are not available on your device, visit myactivity.google.com from any browser. Sign in to your Google account, then click on "Search Services History" in the left sidebar. Uncheck the box next to "Save media." You can also adjust the auto-delete settings here: choose to have saved data removed after 3 months, 18 months, or 36 months. For maximum privacy, select the shortest interval.
Disable Personalized Ads
Even after limiting AI training, Google may still use your media for ad personalization. To stop that, go to myadcenter.google.com and toggle off "Personalized Ads." This ensures your browsing and media history do not influence the ads you see.
Are There Benefits to Leaving It On?
Some users may prefer to keep these features enabled. For instance, if you frequently use visual search to identify products or landmarks, saving your media allows Google to show you your past searches quickly. Similarly, personalized ads can surface deals and products you're genuinely interested in, saving time. Voice search personalization can make speech recognition more accurate for your particular accent or phrasing.
However, these conveniences come at the cost of surrendering your media to Google's training pipeline. The company states that the data is used to improve AI models and services, but the opaque nature of how models are trained means you cannot verify exactly how your data influences the resulting algorithms. For privacy-conscious individuals, the trade-off is rarely worth it.
Other Considerations
This policy change extends beyond just search. Google's broader ecosystem includes YouTube, Google Photos, Google Drive, and Assistant. While the updated language specifically mentions Search Services History, similar data use may occur across other products. Users should review their privacy settings separately for each service. For example, in Google Photos, you can disable "Face grouping" and "Shared library suggestions" to limit how your images are analyzed.
It's also worth noting that opting out today does not retroactively remove previously saved media. Google states that once you disable the setting, new media will not be saved for AI training, but older data may remain in the system until auto-deletion kicks in. For immediate purging, you can manually delete your media history from the My Activity page. Choose "Delete activity by" and select "All time." However, this may also remove search results and personalization you rely on.
As AI continues to integrate into everyday tools, the line between helpful assistance and invasive surveillance grows thinner. Google's latest move is a reminder that default settings often favor the company's interests, not the user's. Taking a few minutes to adjust these preferences can significantly reduce your digital footprint without losing core functionality.
Source: ZDNET News