Skip to content

How we publish, check and correct

Last updated 17 November 2026

CanITrustThat publishes two kinds of work. A teardown is a close study of one app on our test devices. In one teardown, we also studied the device the app controls. A blog post covers a privacy law, a lawsuit, a regulator's action or a security finding that affects mobile apps. Here we set out where our facts come from, how we check each piece, how we use AI, who can open a scan, and how a company or a reader can reply or report an error.

CanITrustThat is run by CITT LLC, a US company. Filip Luchianenco runs it. Every post and teardown is published under his name.

What we publish

We used to publish pages of scan results for single apps, with an app list, a search page and a list called the Hall of Trust. Their addresses were /apps/, /search and /hall-of-trust. We took all of them down on 14 October 2026. The status at those addresses is now "Gone" (HTTP 410).

Scan results

Every scan you start is private, on every plan. That covers rule scans, Deep scans, Ultra-deep scans, questions, uploaded app files and network captures.

  • A scan you start can be opened by your account, by the other accounts on your team or project, and by CanITrustThat staff.
  • A free rule scan (an automated check of an app's files against our rules) can be opened only by your account and by CanITrustThat staff.
  • The fact that you scanned an app is private too.

CanITrustThat publishes no app pages, scores or search results from scans.

Where our facts come from

Teardowns. We decompile the published app. Decompiling turns the finished app back into code a person can read. Then we run the app on test phones and record its network traffic. We install a security certificate on those phones so we can decrypt that traffic (unscramble the parts that are encrypted). Then we can read what the app sent and see which servers received it. Some apps lock certain connections to their own certificate (certificate pinning). For those connections we can see which server received them, but the content is encrypted in the recording. In some teardowns we go further: we watch the app's memory while it runs, or open up the device the app controls. The app version studied and the test setup are stated near the top of each teardown.

Blog posts. Each post is based on the original document: the court filing, the regulator's order or press release, the text of the bill, or the researchers' paper. Those documents are linked in the text. When a fact comes from a news report instead, the news outlet is named and the report is linked.

Code and traffic are two kinds of evidence. An app's code is evidence of what the app contains and can do. A recording of its traffic is evidence of what it sent. In our blog posts and our rule scans, a finding from code states only what the app contains, and a statement that an app sent data needs a recording of its traffic.

How a post is checked before it goes live

CanITrustThat holds editorial responsibility for every blog post. Before a post is published, every link in it is opened and each claim is checked against the source it links to, including for wording that goes further than the evidence. After publication, the live page is checked.

In the Trump Accounts teardown, each code finding was checked again before publication, and each finding rated high or medium severity was tested by an attempt to disprove it.

What we say about named companies

Whether a company broke a law is for a court or a regulator to decide. In our posts we report what a court filing claims, what a ruling or order says, and what we found in the code and the traffic.

  • Allegations are labeled. An allegation is a claim one side makes in a lawsuit, before a court rules on it. When we cover a lawsuit, we label which statements are allegations.
  • The company's answer is included. When a company has answered in public, in a court filing, a statement or a news report, we quote the answer and link to it. When reporters say a company did not answer them, we report that too.
  • Findings have limits. A finding covers one version of an app: what that version contains, or what it sent from our test phones on a given date. The company's reasons for building the app that way, and what the receivers do with the data, are outside the scope of a finding.
  • Device IDs come from our test devices. Device IDs (codes that identify one phone) and similar values shown in a teardown come from our test devices.

How we use AI

AI is part of how we scan apps.

  • Scanning code. In a rule scan, the app's code and files are checked against a fixed set of rules. Each rule is a file with a version number. Each rule file also has a fingerprint (a hash: a short code computed from the file, which changes when the file changes), and the fingerprint of the whole rule set is recorded with every scan. So the same app file and the same rules give the same rule findings. In a Deep scan and in answers to questions, AI models also read the code. The code files the model reads, and the text of any question, are sent to the AI provider named in our privacy notice. When a model answers a question about an app, its findings are labeled as AI output, and code checks that each file it cites exists in the app. A finding that fails the check is dropped. Rule scan findings come from the rules alone.
  • Running apps on test phones. The app is operated automatically on our test phones while its network traffic is recorded. Screenshots of the tested app may be sent to the AI providers named in our privacy notice.
  • Summing up a recording. In the summary of a recording of an app's traffic, the counts and the wording come from fixed code, which gives the same output every time.

Money and independence

Findings come from an app's code, store listing and privacy policy, and from its network traffic when we record it; we earn money from paid plans and commissioned work, we accept donations, and no payment changes a finding, a score, a label or a published page.

Scans and client projects are private, as set out under Scan results.

Replies from companies and developers

A company or developer named in a post or teardown can reply by email to [email protected]. The email should come from an address at the company's own web domain (for a company at example.com, an address ending in @example.com), or from the contact address on the app's store listing. It should name the post.

We add the reply to the end of the post or teardown, with its date. Before we add it, we remove abuse and personal data. A developer can also ask us to look at a newer version of the app.

Corrections and updates

Anyone can report an error by email to [email protected]. The email should name the post, the sentence, and the source that shows the error.

When we correct a post or teardown, the corrected piece shows the date of the correction and what changed. The date of the last update is shown at the top of each blog post and teardown.

When we take down a post

We take down a post or teardown, or remove part of it, only when:

  • the wrong app or company is named in it;
  • it contains a customer's private work, or personal data about a private person found inside an app;
  • a court or a government authority orders it.

When someone asks us to take down a post or teardown for any other reason, we offer the right of reply instead. That is the chance to have a reply added to the piece, as described above.