Google's default AI training on Search, Lens, and Translate puts Australian businesses at risk

Google's latest change goes beyond a typical AI update. The company now uses saved media—screenshots, photos, audio, and video—from everyday interactions with Search, Lens, and Translate to train its models unless users manually opt out. This isn't public web crawling; it's the error message an employee screenshotted or the supplier document they translated. With Google holding roughly 88% of Australia's search market (over 90% on mobile), the default setting captures data from millions of users who never signed up for AI training.

The timing is critical. Australia's Privacy Act will require entities to disclose how personal information is used in automated decision-making from December 2026. Businesses must now treat a routine Lens search as a potential data governance issue: that photo of a faulty product or contract term could become AI training material. IT teams can't assume staff have opted out, and disabling the setting doesn't delete already-saved data. This mirrors Meta's 2024 controversy where it scraped Australian Facebook and Instagram posts without local opt-out. For organisations, the risk is that a consumer privacy choice quietly becomes a corporate compliance headache.