Is your phone listening to you for ads?

You talk about getting your child a new school bag. Later, you open Facebook and there it is. A school bag. Close enough to what you were discussing that the phone on the kitchen table suddenly feels less like an object you own.

It is tempting to treat that as the whole explanation: it heard us.

But the ad doesn't show you how it got there. You may have visited a shop, followed a parenting account, bought something from a business that shares customer information, or simply fallen into a group of people being shown school bags that week. You cannot work backward from that one ad and know which happened.

So there are two things to check. Whether an app accessed your microphone, and what information it can get without ever hearing the room. The second can tell a company quite a lot about your family on its own.

The “Active Listening” story had a later chapter

One of the more alarming stories involved companies selling an advertising service called Active Listening. Their pitch described using conversations captured by smart devices to find customers. That sounds like confirmation of exactly what people feared.

Then the Federal Trade Commission investigated. In August 2026, it finalized settlements with Cox Media Group and two other companies over allegations that the service was falsely advertised. According to the FTC, it wasn't using voice data, and consumers had not opted in as claimed. The combined settlements required $930,000 in payments. The FTC's final announcement.

A sales pitch for listening turned out not to be evidence that listening had happened. That doesn't clear every app of every possible misuse. It does mean this particular story needs its ending attached when someone uses it to explain the ad on your phone.

What Facebook says, and what Meta AI changes

Meta says Facebook uses the microphone when you have given permission and are actively using a feature that needs it. That is the company's account of its behavior; the school-bag ad is not evidence that contradicts it. Meta's microphone explanation.

There is a separate, documented use of conversations worth knowing about. Meta announced that, from December 16, 2025 in most regions, text and voice interactions with Meta AI would help personalize content and ads. It names sensitive topics, including health and religion, that it says it does not use for ad targeting. The announced change.

Speaking deliberately to an AI assistant and chatting with your family beside an idle phone are different situations. If you ask an assistant for help with your child's routine, you have given it the information in your question. You don't need to include a full name, school, address or identifying photograph for it to understand a general question.

And if the microphone is active, consider what is happening behind you. A child's voice can be part of a voice request even when the child wasn't the person asking. The smart-glasses guide covers Meta's separate policy for recordings and background sound on its glasses.

An ad can follow information you never posted

Facebook's advertising explanations describe information from activity on its services and from other businesses. A shop can also match a customer list to accounts using contact details. Neither requires you to make a public post about what you bought. Meta's explanation of advertiser lists.

Think about an adult buying children's clothes, looking at clubs, or registering for a family event. Those actions can involve information about children even when the children have no accounts. This is why “our child isn't on Facebook” and “Facebook receives nothing about our family” are different claims.

When an ad feels unusually specific, open its three-dot menu, then Why am I seeing this ad? Facebook provides an explanation and a route to Ad preferences. It may name activity or interests that help make sense of the ad. It is a summary of targeting factors, not a recording log or proof of one precise cause. How Meta describes the explanation.

For the controls behind those factors, use the Facebook data-settings walkthrough. Turning off the microphone and limiting outside-business personalization are separate jobs.

Check the microphone on your phone

Start with the app you are concerned about. You don't need to test it by saying an unusual word out loud and waiting for an ad; a result from that experiment still won't tell you where the targeting information came from.

On iPhone

1. Open Settings → Privacy & Security → Microphone. This shows which apps have permission. Switch it off for an app that doesn't need to record sound. Apple's permission controls.
2. To see access over time, open Settings → Privacy & Security → App Privacy Report and turn it on. On iOS 15.2 or later, Data & Sensor Access shows the preceding seven days, but the report only starts collecting after you enable it. Come back after using the phone to see the entries. Apple's report.

An orange indicator means the microphone is being used; green means the camera, or camera and microphone together. Apple's indicator guide.

On Android

1. Open Settings → Security & privacy or Privacy → Privacy dashboard → Microphone. You may need to open Privacy again before the dashboard. Android 12 shows access over 24 hours; Android 13 shows seven days. Google's dashboard instructions.
2. Tap a listed app to change its permission. You can also open Settings → Apps → [app name] → Permissions → Microphone → Don't allow. Google's permission instructions.

If you find access you didn't expect

Compare an access time with what you were doing. A voice message or video call explains microphone use. An entry you cannot account for deserves a closer look: note the app and time, revoke the permission if it isn't needed, update the app and phone, and check again. If unexpected access continues, contact the app's support with those details.

A permission being enabled does not mean the app recorded continuously. A microphone-access entry does not tell you what was said or whether the sound became advertising data. And an empty report cannot tell you what happened before its recording window began. Use the report to answer the question it can answer: which app accessed this sensor, and when?

Protect the conversation as well as the photograph

A child's face can be covered throughout a clip while someone says their name in the background. Microphone settings won't remove words from a file you've already recorded.

Before posting a video or sharing a screen recording, listen to what the file contains. If it includes a child's name, school or something personal said in the background, remove that sound before sharing. For a live AI conversation, move away from private conversations before starting, or type the question instead.

Then make the phone changes that match what you found. An unnecessary microphone permission can be removed. Outside-business personalization has its own control. An identifying detail can be left out of the next AI question. You can make these changes without proving that the school-bag ad came from the kitchen conversation.