Uh-oh, Google trains search AI using web content despite opt-outs

Google‘s latest court testimony reveals a significant loophole in its AI training opt-out system, potentially undermining publisher control over how their content is used. This disclosure highlights growing tensions between tech giants and content creators as AI systems increasingly rely on web content for training while offering inconsistent protections for publishers trying to maintain rights over their intellectual property.

The big picture: Google’s AI training controls allow publishers to opt out of having their content used for AI development, but this protection only applies to Google DeepMind‘s work, not other AI products within the company.

Key details: Eli Collins, a Google DeepMind vice president, testified in court that Google can train its search-specific AI products like AI Overviews using content from publishers who have explicitly opted out of AI training.

• The testimony clarifies that Google’s opt-out controls only cover work done by Google DeepMind, the company’s AI research lab.
• This creates a significant distinction between how different Google divisions handle publisher content preferences.

Why this matters: This revelation exposes a critical gap between what publishers believe they’re protecting when they opt out of AI training and what’s actually protected under Google’s current system.

• Publishers seeking to prevent their content from being used to train AI may not realize their work could still be incorporated into Google’s search-specific AI products.
• The distinction between different Google divisions creates a complex landscape for content creators trying to control how their intellectual property is used.

Reading between the lines: Google’s internal organizational boundaries are creating policy inconsistencies that could undermine trust with publishers and potentially attract regulatory scrutiny.

• By maintaining separate policies for different AI initiatives within the company, Google effectively circumvents publisher preferences through organizational structure rather than technical limitations.