Google's privacy policy now lets Gmail data shape ads if you link it to its AI tools

For years, Google has said it doesn't use what's inside Gmail, Drive and Google Photos to pick the ads you see. On Oct. 16, that promise gained a condition.

The revised Google privacy policy now says personalized ads won't draw on content from those three services "unless you choose to make data from these services available" to other Google products that support personalized ads. Android Authority spotted the edit first. No announcement came with it. That's the part I keep coming back to. A pledge this old deserved a louder funeral.

The Google Privacy Policy Exception Runs Through AI

The condition is tied to Personal Intelligence, Google's effort to let AI Mode in Search and the Gemini assistant read a person's own connected apps. Link your inbox and the tools can do useful things, like dig up a hotel reservation you'd forgotten about. I'll grant that it's useful.

But under the new wording, that same connected data can also inform which ads you're shown. One switch, two consequences, and only one of them is advertised on the toggle people think they're flipping.

Google has set some limits, according to the Decrypt report. Nothing changes by default: ad targeting from Gmail, Drive or Photos content stays off until a user connects the data. Advertisers don't see the contents of anyone's email, files or pictures. The company also says it won't hand over personally identifying details without permission.

So this isn't a return to the days when Gmail messages were scanned for ad keywords. Google ended that practice years ago. The change is narrower. It is still a retreat from an absolute promise, and absolute promises are easy to explain. Conditional ones are not.

Opting Out Is Possible, Forgetting Is Harder

People who've already connected their apps can back out. Go to the Connected Content apps section under Data & privacy in your Google Account, or turn off personalized recommendations. One caveat from Android Authority: information gathered earlier may still sit in your Search history after you disconnect.

That gap matters. Disconnecting stops future use. It doesn't obviously erase what the system already learned.

There's a pattern here, too. In May, Decrypt reported that Chrome had begun quietly putting a multi‑gigabyte AI model onto people's machines, where it runs locally. The same report said Google had removed a privacy assurance it once attached to Gemini Nano. Different products, same rhythm: the AI ambitions go first and the explanation trails behind.

Why Advertising Is the Pressure Point

The business logic isn't mysterious. Search advertising is still the engine of Alphabet's revenue, and AI answers threaten to loosen it. When a chatbot replies directly, fewer people click through a page of sponsored links. Advertisers pay for intent, and a conversational answer can satisfy intent without a click.

Personal data is the richest fuel available for making AI responses and ads feel specific. A flight confirmation in an inbox or a trip album in Photos says more about what someone may want next than a single search query does. Connecting the two was always going to be tempting.

That is also why the wording matters to regulators. In Europe, consent has to be specific and informed, and a single toggle that bundles a convenience feature with an ad‑targeting permission invites scrutiny over whether it is either. In the US, the Federal Trade Commission has long treated a gap between what a company says and what it does as a deceptive practice. Neither has to act for the question to hang over the rollout.

What I'd Watch Next

I'd watch the consent screen. If the prompt for linking Gmail to Gemini spells out the ad consequence in plain language, I'd call this a defensible tradeoff. If it's buried in a policy page few people open, regulators in Europe and the US will have an easy question to ask. Neither Google nor any regulator has said yet which way this is heading.