Where does DeepSearch get its data?

DeepSearch gets its data from public web pages, social profiles, public records and news, from licensed numbering and people-data services, and from public breach-notification indexes. Which of those a search uses depends on the lookup and the product surface.

The category matters more than the list. A carrier label describes a number, a public profile describes an account, and a broker record describes an association someone else compiled. None of them alone verifies the person behind it. Our privacy policy and terms are the binding text for sources, access, retention and removal.

What data sources does DeepSearch use?

DeepSearch can use seven source categories, depending on the lookup and product surface:

• public web pages
• public social media profiles
• public records and directories
• news and published articles
• licensed telecom carrier and numbering services that identify a number's carrier, line type, and assignment region
• public breach-notification indexes that name incidents associated with an email address
• licensed people-data and identity-resolution providers, on the account-based surfaces that offer them

None of these categories is secret. The FTC describes the whole people-search category the same way: such sites may compile government public records, collect public social media profiles and buy information from other data brokers. What a reader needs is to know which category a given claim came from, because the category sets how much weight the claim can carry.

Licensed people-data results are limited to signed-in accounts. Our MPP research documentation, which covers the anonymous pay-per-report API, says people search and licensed data brokers need a DeepSearch account, "where results keep their provider attribution."

Paying for a dataset does not make its identity associations verified facts, and none of these categories provides private-account access or leaked credential contents. The guide to what each source category can and cannot establish takes them one at a time.

Read the source category

Access to information is not verification of a person

The available sources depend on the lookup and product surface.

Public sources

• Webpages, profiles, public records and news.

  Check the original page, its date and what it says.

• Public breach-notification indexes.

  An incident association, not access to leaked contents.

Licensed sources

• Carrier, line type and numbering metadata.

  Describes the number; does not verify its current user.

• People-data and identity-resolution providers.

  Keep provider attribution; an association is still a candidate.

Outside the boundary

Consumer service

• Private accounts and messages
• Passwords and leaked credentials
• Live location
• Private carrier subscriber records

Can DeepSearch access private accounts or messages?

No. The privacy policy sets the boundary for the consumer service: "We do not access private accounts, private messages, passwords, leaked credentials, live location, or private carrier subscriber records." Each exclusion has a cost.

No leaked credential contents. A public breach-notification index may tell us the name of an incident associated with an email address. DeepSearch does not provide passwords, hashes, authentication codes or the contents of a breached record. The cost is that an email check can name an incident, but not say what it exposed about you.

No private accounts. We don't access private or follower-only accounts. If a profile is private, it stays private, and the terms forbid anyone from using DeepSearch to "discover or expose private accounts."

No call or message contents. DeepSearch does not provide private inboxes, direct messages, call audio, text-message contents, or call history. Carrier call records also have federal protection: under 18 U.S.C. 1039, obtaining them by fraud, or selling them without the customer's prior authorization, carries up to 10 years in prison.

No private carrier subscriber records. Licensed numbering data can identify a carrier, line type, and assignment region, and in limited cases it carries a caller label. It does not reveal communications or show the device's location, and a caller label is not a verified subscriber identity.

No live location. We can't tell you where someone is. A public post can show where someone was when it was published, and photo location estimates where an image was taken. Neither is access to a person's device.

No identification from a face. A reverse-image search looks for the same or a similar image on public pages. It finds where a picture appears; it does not identify the person in it.

How can you check the source of a result?

Open the link behind the claim. Every public-web fact in a profile should carry a link to the page it came from, and licensed numbering, people-data and breach-index fields should be labeled by source category or provider. If our summary says someone is a designer in Austin, you should be able to click through, see the page that says so, and decide whether that page is credible.

The links also make our mistakes visible. A sentence with no source behind it, or a source that doesn't say what the summary says, is something you can catch and report. The terms hold developers to the same rule: API customers may not "remove source attribution or safety notices." The guide to checking a result against its sources breaks the check into four questions: retrieval, attribution, interpretation and freshness.

Example: reading the sources in a phone and email lookup

Illustrative example, not a real search result. You are about to buy a used camera from a marketplace seller who signs messages "Robin Vale" and gave you the email robin.vale@example.com and the number (202) 555-0147. You run an email search and a phone search before paying. A result might contain these fields.

A good summary of that result says the handle and email connect the marketplace profile and the forum post, and that no source in this search ties the phone number to the name. The breach entry says nothing about whether the seller is honest, and no field in the result can show the seller's live location, messages or character.

Does DeepSearch generate information with AI?

DeepSearch uses AI to work with what a search found, not to supply facts of its own. The privacy policy lists what it may do: interpret a prompt, generate search variants, rank possible matches, extract facts, summarize sources, translate text, or draft an answer.

That means the AI layer should say "no source in this search ties this person to that address" rather than guessing, and "possible match" rather than asserting an identity the evidence doesn't support. Models can still get this wrong, so the policies treat the output as a draft.

The privacy policy says AI output "can be inaccurate, incomplete, outdated, or attributed to the wrong person" and must not be treated as verified fact or used for a legally significant decision. The terms tell users not to represent it as verified by DeepSearch.

Can sources about different people be mixed up?

Yes. Records about different people who share a name can be combined, so a name alone never establishes a match and a single signal stays a hypothesis until independent sources agree. Candidates are ranked with a match score, which our terms say means only that "the system found signals consistent with a candidate."

The cost is that you may see two candidates where you hoped for one. A confident wrong profile about a real person is worse than an incomplete one. The longer note explains why a shared name is never treated as proof.

Does DeepSearch store your searches and reports?

Yes, in two different ways. The privacy policy treats searches, prompts, filters, candidate choices, reports opened and chat threads as usage data DeepSearch may collect. Separately, the person, company and vehicle reports DeepSearch builds are stored by subject in a shared cache.

The shared cache is the part most readers do not expect. A report built for one customer can be shown to another who later looks up the same subject, and the cached report does not record who requested it. Deleting your account does not remove cached reports about other people. The person a report is about can suppress it, as the next section describes.

Your own account content is generally kept while the account is active, then deleted or anonymized through account deletion, subject to exceptions such as billing, security and legal records. Account deletion and Remove my info are separate actions: one closes your account, the other stops DeepSearch showing a result about you.

On selling data, the policy separates two things. We don't sell the payment credentials, private account content or Data Cleanup profile (the details you give our paid broker-removal service) for money. We do run a paid service whose core function is to "find, organize, and provide access to public and licensed information about search subjects," and some state laws may count that paid access as a sale.

How do you remove your information from DeepSearch?

If you appear in DeepSearch results, you can remove yourself for free and without an account on Remove my info. Give your name and an email address you can open, and enter the code we send. We then withhold every identifier on the provider record that address matches, so the removal keeps working when the provider refreshes its data. It applies on the website, the apps and the API.

The same page takes down our public phone-number and public-figure pages. A phone number or direct page link is normally suppressed as soon as the form succeeds. A name alone is held for review, because a name-wide block could remove a different person's page; we aim to finish that review within 48 hours.

To correct a result instead, email support@deepsearch.app with the page URL, the field, why it is wrong, and a reliable public source. We will not ask you to prove that every true public fact should remain accessible.

Appeals go by email with the subject "Removal appeal". An urgent safety risk goes to support with "Urgent safety" and the exact URL; that is not an emergency service.

Our removal covers our output. It cannot delete a court, government, news or data-broker record, or a search engine's cached copy. Remove my info is also separate from Data Cleanup, which sends opt-out requests to participating data brokers. The guide to which correction or removal route fits the problem sorts the cases, and the four-layers note explains what each removal route can and cannot reach.

What can DeepSearch results be used for?

Personal research you then verify at the source. Some of the limits are law and some are our own line; both are in our terms.

Not for FCRA-covered decisions. We are not a consumer reporting agency. Results must not be used to decide eligibility for employment, credit, insurance, housing or tenancy, education, government benefits or licensing. Those decisions require a screening provider operating under the Fair Credit Reporting Act, which allows a report only for the permissible purposes in 15 U.S.C. 1681b.

Under the FTC's guidance for employers, an employer also needs written permission and, before any adverse action, must give the applicant a copy of the report and a summary of their rights.

We provide none of that. The terms add that obtaining a person's consent "does not convert DeepSearch into a compliant screening product." The longer note covers what separates a people-search result from a consumer report.

Not for stalking or harassment. Aggregating public information to monitor, locate, or pressure someone who doesn't want contact is the harm this category is criticized for, and the criticism is fair. The information being public doesn't make the aggregation harmless.

Not for doxxing. Publishing someone's home address or workplace to expose them is not research.

Not for face identification or tracking. The terms forbid building "unlawful surveillance, facial-recognition, biometric-identification, or tracking systems" with DeepSearch.

How the rules are enforced. The terms make the user responsible for the purpose of a search. They let us require "email confirmation, reauthentication, payment verification, or other proportionate checks," and let us restrict or end access when we reasonably believe there is misuse.

Where can you check DeepSearch's data policies?

Each commitment above is written in a public policy with its own "Last updated" date, so you can check us against the current text.

If one of those pages changes, this post should change with it. If you find a result that crosses one of these lines, or find us falling short of any of it, keep the result URL and the source and tell us. The contact page reaches us.