The AI Boom Runs on Data Brokers

Everyone pictures AI learning from the open internet. Wikipedia, news sites, the public web. That is the part you can see. The part you cannot see is an older industry that has spent decades quietly buying and selling the details of your life, and that has suddenly found itself sitting on the most valuable inventory in tech.

Data brokers are not new. AI just made them rich.

The supply chain behind the models

A data broker collects personal information, combines it, and sells access to it. The raw material comes from everywhere. Mobile apps, loyalty programs, public records, online activity, and lists bought from other brokers. You never opened an account with them. There was no consent screen and no clear way to opt out, yet they hold a profile of where you live, what you buy, who you are related to, and how you tend to behave.

For years that profile was sold to advertisers, insurers, and lenders. Then AI arrived with a new and bottomless appetite. Public text alone does not make a model good at understanding people. Detailed, real-world behavioral data does. So a new line of business opened up, and brokers started packaging curated datasets aimed squarely at training and tuning AI systems.

What you can do, and what it will not fix

You have more leverage here than with a finished model, because the broker supply is ongoing rather than one-and-done.

What helps:

 
• Use your state deletion rights where they exist. Several states now require brokers to honor deletion requests, and a few are building single-request systems to make that easier.
 
• Opt out broker by broker. It works, but each site has its own process and most will quietly relist you.
 
• Cut the supply. Removing yourself from the brokers stops them from continuing to package and resell a fresh copy of you.

What it will not fix: deleting from a broker today does not retrieve data that has already been sold, and it cannot reach into a model that already trained on it. What it does accomplish is turning off the tap. A profile that stops being updated stops being worth much, and that is the part you can still control.

Brokers are betting that the cleanup is too tedious to finish, and the evidence is on their side. Consumer Reports measured automated and self-service removal clearing about 27 percent of listings, against roughly 70 percent for removal carried out by people. Persistence is the entire product, and persistence is exactly what most people run out of around week two.

Frequently asked questions