Motivated intruder test explained with a real example

The motivated intruder test decides whether data remains personal after you strip obvious identifiers. If a determined person could still link the data to an individual, the data is not de-identified.

What the law says

• HIPAA Safe Harbor. Remove 18 specific identifiers such as names, exact dates, and fine-grained geographic details. Once those are gone, the information is no longer "protected health information."
• GDPR Recital 26. Personal data includes any information that can be linked to a person, directly or indirectly. If a reasonable person cannot identify the individual, the data is considered anonymous.
• CCPA §1798.140. The "reasonable person" test asks whether an ordinary consumer could discover a person's identity from the data. If not, the data falls outside the act's privacy protections.
• Australia's Privacy Act (APP 6). Allows de-identification when the data cannot be used to identify an individual, either alone or when combined with other information.

Each rule boils down to one step: remove anything that could allow someone to reconstruct a person's identity, and the data loses its personal status.

How the test works

Picture a 70-person manufacturing firm that wants to share ticket-resolution logs with an AI lab. A raw ticket might look like this:

Ticket #3421 to 2023-07-15
Customer: Acme Corp
Issue: Machine #5 stopped at 09:23 due to sensor error 0xA7.
Resolution: Rebooted controller, replaced sensor, tested for 30 min.
Handled by: Jane D.

First. Delete the ticket number, exact dates, the customer name, and the employee name. The record becomes:

Ticket, 2023-07-??
Customer: redacted
Issue: Machine #5 stopped at ?? due to sensor error 0xA7.
Resolution: redacted
Handled by: redacted

Second. Ask the motivated intruder question: could someone who knows your factory layout, typical sensor-error timing, and the roster of technicians pinpoint which employee wrote the ticket? If you also replace the machine number with a generic label and hide the exact time, the answer is likely no. The record then passes the test and is treated as de-identified.

The test is a concrete checkpoint, not a vague guess. It forces you to consider every element, dates, locations, device IDs, job titles, that could combine to reveal an identity. When the test is satisfied, you have a clear legal footing to share the data with a buyer or partner.

Need data de-identified to meet HIPAA, CCPA or GDPR? See how we do it

Why the test matters

Regulators do not accept a "good enough" guess. The test provides a legal checkpoint that tells you whether you can safely license, share, or retain the data without breaching privacy law. It also reassures AI labs that the data they receive does not expose you to liability for mishandling personal information. Because a trained model cannot reproduce the original records, use rights are usually perpetual, while exclusivity periods are limited, keeping the arrangement attractive to both sides.

Our de-identification service

We guide you through the motivated intruder test and perform the necessary scrubbing. First, we map the systems that hold the data and identify which laws apply. Next, we export raw files to a secure, logged workspace and delete the originals on a recorded date. Then we replace names, IDs, dates, locations and other identifiers with stable placeholders such as "employee 632." Finally, we produce a written record of the cleaning process and the test results. The de-identification occurs before any AI lab sees a sample, and your counsel confirms that the process satisfies GDPR, CCPA, HIPAA, the UK ICO, or Canada's PIPEDA. Read more on our de-identification page and see the regional breakdown on our regions page.

In most licensing deals the AI lab pays for the de-identification work, but the exact arrangement is negotiated case-by-case. Some sellers prefer to have the data cleaned first, which can lead to a stronger offer. The contract will specify who handles the cleaning, how raw files are treated, and when they are destroyed.

When to use it

• Preparing data for AI labs. Before you list your operational data on the sell page. You need proof that the data is no longer personal.
• Sharing data with partners. If a partner requests a data set, run the test and provide the de-identified version.
• Internal compliance checks. Embed the test in your routine privacy audit to ensure new data collections follow the same rules.

How to request a quote

If you want a formal quote for our de-identification service, let us run the test on your data and produce a compliance memo. Click the button below. We will ask a few basic questions, keep your data secure, and return a clear quote that fits your situation.