How to Opt Out of Google Search’s New AI Data Training Feature

The more telling detail is what Google is collecting: visual media uploaded during Search interactions, not just query text. That represents a qualitative expansion of training data diversity, and it arrives precisely when the industry is hunting for non-synthetic, real-world multimodal content to push model capability forward.

The data grab fits a pattern of infrastructure-layer competition that has been building across the coverage period. While the Cerebras earnings story from June 24 illustrates how brutal the economics of AI hardware have become, Google's move here is a reminder that compute is only one input constraint. Training data, particularly diverse visual content at scale, is the other. Google is essentially monetizing its search surface as a data pipeline at a moment when rivals without comparable consumer touchpoints face real scarcity. The opt-out default also signals that Google is willing to absorb regulatory friction to close that gap faster.

Watch whether the EU's data protection authorities issue a formal inquiry within the next 90 days, since GDPR's opt-in requirements for training data use would force a structural reversal of this default in European markets, which would meaningfully limit the volume and geographic diversity of what Google actually collects.