New to Rust? Grab our free Rust for Beginners eBook Get it free →
What We Can Learn About the Didi Data Breaches

An app team reviewing what its product collects may want a concrete case to compare against its own controls. The 2022 Didi decision by China’s Cyberspace Administration (CAC) shows what regulators found about data collection and handling, and what those findings can teach product teams.
This article separates the documented violations from what the public record does not establish, then turns the findings into checks an app team can use.
TL;DR
The 2022 Didi decision documented extensive unlawful data handling, but the public findings do not describe an outside attacker stealing a database.
- The Cyberspace Administration of China (CAC) fined Didi RMB 8.026 billion on July 21, 2022.
- The CAC listed unlawful collection and other data-handling violations.
- The regulator’s counts refer to pieces of information, not a count of unique people.
- For product teams, it highlights data minimization and evidence that access controls work.
What Happened in the Didi Data Case?
The Didi case was a regulatory investigation that ended with the CAC finding unlawful data-handling practices and fining the company RMB 8.026 billion. Keep the original RMB amount in summaries because dollar conversions vary with exchange rates and publication dates.
| Milestone | What the CAC published |
|---|---|
| Earliest conduct | The CAC Q&A dates it to June 2015. |
| July 2021 review | The CAC announced a cybersecurity review of Didi. New-user registration was restricted while it was underway. |
| Penalty decision | The CAC penalty announcement cited the Cybersecurity Law, Data Security Law, and Personal Information Protection Law. |
| Company and executive penalties | Didi was fined RMB 8.026 billion. Chair and CEO Cheng Wei and president Liu Qing were each fined RMB 1 million. |
The notice states the CAC’s legal basis. A different company should get qualified advice on its own obligations.
Does a Data-Handling Violation Mean a Hacker Breach?
The FBI Internet Crime Complaint Center defines a data breach as an intrusion resulting in unauthorized access to protected information. Legal wording can vary by jurisdiction, so name the source when using the term.
The CAC’s public findings concern excessive collection and unclear purposes, without identifying an outside attacker or database theft.
| Situation | What it establishes | What it does not establish by itself |
|---|---|---|
| An outside party breaks into a service and accesses stored records | An intrusion and unauthorized access, if the investigation confirms them | Which people were affected until the service verifies the scope |
| An app collects more information than its stated task requires | A collection or privacy-control problem that needs investigation and correction | That an outside attacker accessed or copied the collected information |
| A service stores sensitive information without adequate protection | A security exposure that can raise the consequences of future access | That an unauthorized party has already obtained the information |
How a Breach Response Progresses
The FTC business response guide offers a general sequence for handling a suspected incident.
- Detect: Flag unusual access or transfer for review.
- Validate: Check whether access was authorized and identify the information involved.
- Contain: Stop ongoing access while preserving relevant evidence.
- Recover: Restore normal operations after the team checks the affected controls.
- Communicate: Report verified facts and required next steps under applicable rules.
Preserve access logs before changing permissions because they can establish which role read a field and when.
What Did the CAC Say Didi Collected or Processed?
The CAC said Didi collected or processed phone, passenger, location, identity and travel-intent information, including data it found unnecessary or insufficiently disclosed. Its Q&A grouped 16 findings into eight categories.
| Finding in the CAC Q&A | Reported quantity or example | How to read it |
|---|---|---|
| Screenshot information from phone photo albums | 11.9639 million pieces | The quantity reported by the regulator. |
| Clipboard and installed-application-list information | 8.323 billion pieces collected beyond what was needed | Two classes of device information were described as over-collected. |
| Passenger profile information | 107 million facial-recognition items 53.5092 million age-group items 16.3356 million occupation items 1.3829 million family-relationship items | The regulator listed separate quantities for these categories. |
| Home and work ride-address information | 153 million pieces | Trip destinations can identify routines when linked to a person. |
| Precise location information | 167 million pieces | The Q&A named situations involving app background operation and connected vehicle equipment. |
| Driver education and identity information | 142,900 education entries and 57.8026 million ID numbers reportedly stored in plaintext | The CAC’s listed issue is plaintext storage of identity numbers. |
| Passenger travel-intent analysis without clear notice | 53.976 billion travel-intent items 1.538 billion resident-city items 304 million non-local business or travel items | The Q&A flags analysis without clear disclosure to passengers. |
| Irrelevant phone permissions and unclear processing purposes | Frequent requests for unrelated phone permissions and unclear purposes for 19 information types | Permission prompts and privacy explanations must match the feature that needs the data. |
Describe the 64.709 billion as information items because the CAC reports pieces or entries, not unique people or accounts.
Review the full data path because the findings include issues beyond permission prompts.
Some examples help explain why collection boundaries matter. Precise coordinates can describe where a ride began or ended, and home or work labels can reveal repeated destinations when connected to a rider. An identity number stored in plaintext can be read directly by a party that gains access to the relevant storage.
The Q&A also mentions frequent requests for irrelevant phone permissions and unclear purposes for 19 information types. Those findings call for separate checks: when a device capability is requested and whether the notice explains later use.
What Can Product Teams Learn From the Didi Case?
Product teams can turn the CAC’s findings into release tests that match each data path.
| Decision | Control to review | Evidence a team can keep |
|---|---|---|
| Does the feature need this field? | Remove collection that does not support a defined user task, and revisit that decision when the feature changes. | A data inventory with owner, purpose, source, and retention rule. |
| Can the user understand why it is collected? | Use a clear notice at the point where the information is requested or inferred. | Versioned notice text linked to the feature and release. |
| Does the permission match the task? | Request a device permission only when the feature needs it, and explain the consequence of declining. | Permission tests for first use, denial, and later settings changes. |
| Can staff or services read sensitive fields? | Limit access to named roles and services, and protect identity data at rest. | Access reviews, encryption settings, and a record of changes. |
| Can the team prove its controls work? | Test collection and deletion paths during release and after material changes. | Test results, issue ownership, remediation dates, and a retest. |
Review the permission flow alongside notice text because the dialog controls device access, while the notice should explain the product’s purpose.
Test the permission flow against network requests. A feature should send only required fields when access is granted and use its documented fallback when access is denied. On a versioned build, trace a test field into analytics and diagnostic records, then check storage and read access with synthetic data.
Consider a feature that needs location only at pickup. Capture a staged request while permission is granted, then inspect the analytics event after the task ends for a retained coordinate. If the product needs only a city-level trend, store a coarser output and keep the results with the build identifier.
CodeForGeek’s web development cybersecurity guide covers request handling and access checks. Its code-scanning tools comparison helps teams ask where source code goes during a hosted pilot. These checks cannot guarantee an intrusion will not occur, but they can reduce unnecessary collection and help a response team explain a control failure.
What Should You Do After a Data-Breach Notice?
After a data-breach notice, verify it through the service’s known website or phone number, then take the steps that match the data involved. Ask which data category is covered because that detail guides the response.
This table is general guidance for a notice from a service you use, not a claim that Didi sent an individual notice about its 2022 decision.
| What the notice says | Next move | Why it helps |
|---|---|---|
| Login credentials may have been accessed | Change that password on the affected service and anywhere else you reused it. Enable multi-factor authentication when offered. | A reused credential can open other accounts. |
| Payment or identity data may be involved | Contact the relevant bank or issuer using a known channel and review account activity. | The response can be matched to the type of information named. |
| The notice is vague or unexpected | Verify it through the service’s known website or phone number before clicking links or sharing details. | Scammers can imitate breach notices to collect passwords or payment data. |
| You manage the affected service | Preserve evidence, determine the affected systems and information, contain further access, and follow the notice rules that apply in your jurisdiction. | The response needs verified scope and applicable legal guidance. |
For a US-oriented response, the FTC guide recommends preserving evidence, while FBI IC3 covers containment and credential resets after a confirmed intrusion.
Save any reply with the notice and record the date received.
A rider can open the service through a known address. Official support can explain its privacy controls and account-data request path. These steps address one account, while the CAC’s published quantities remain aggregate.
What the Public Record Does Not Establish
The public record does not establish every permission request’s app version or screen, a current technical audit of Didi’s apps, or the details the CAC withheld. The CAC cited national-security reasons for withholding some processing information.
- The public Q&A does not map every permission request to a specific app version or screen.
- The public documents do not provide a current technical audit of Didi’s apps.
A reader comparing this case with another incident should check that the event type and jurisdiction match. US notification records and Chinese administrative findings describe different processes.
A product team can turn each published category into a control owner and retest date because the case summary does not assign work for another company’s next release.
When citing the case, keep the CAC’s original page beside the English summary so another reviewer can check translated terms.
Conclusion: Measure What the App Collects
Use a dated copy of the CAC record when teaching the case.
Compare a field-level inventory against the controls above, then read the CAC’s Q&A for the source findings.
Additional Resources
Start with the CAC notices for the Didi findings because they provide the primary case record. Use the other links for general definitions and response guidance.
- CAC Q&A: the regulator’s public explanation of its findings and investigation.
- CAC penalty announcement: the decision, statutory basis, and penalties.
- FBI IC3 data-breach page: a US-oriented explanation of the term.
- FTC response guide: business steps for investigating a suspected breach.
- Privacy Rights Clearinghouse chronology: public notices from US cases.
FAQ
These answers cover questions that remain after reading the public decision.
Why did the CAC withhold some details?
The CAC said some data-processing details related to national security and were not disclosed. The public record does not support guessing what those details were.
Do the CAC’s quantities show how many people were affected?
No. The decision counts pieces or entries of information in its findings. It does not provide a verified count of unique people or accounts represented by those items.
Why can ride-location data be sensitive?
Precise trip coordinates and repeated home or work destinations can reveal a person’s routines when linked to an account. The CAC named location and home/work ride-address information among its findings.
Did the CAC say hackers stole Didi’s database?
The public CAC decision and Q&A describe data-handling findings. They do not name an outside attacker or report database theft, so a hacker breach is not established by those documents.
What should an app team review first?
Map one feature’s data field to its purpose and owner. Check the permission path on a test build. Use a synthetic record to verify how a sensitive field is stored and later removed.
How can a rider check whether their account was affected?
The public CAC quantities are aggregate information counts, not account-level records. A rider needs a service-specific notice or answer from official support to learn about their own account.
What does data breach mean in this article?
The article uses the FBI IC3 definition as a US-oriented explainer: an intrusion that results in unauthorized access to protected information. Other jurisdictions may use different wording.
How should a developer apply the findings?
Use the table as an audit prompt rather than a reproduction of Didi’s app. Follow a feature’s requested data through an analytics copy, and test permission and deletion paths on a versioned build.
When did the CAC review and penalty occur?
The CAC announced its cybersecurity review in July 2021. Its administrative penalty decision followed in July 2022. The Q&A dates the earliest conduct to June 2015, a separate milestone.
What fine did the CAC impose?
The public announcement set Didi’s fine at RMB 8.026 billion. It also fined two executives RMB 1 million each. Those are regulatory penalties, not per-user compensation amounts.
How should I report the 64.709 billion figure?
Describe it as an information-item total because the CAC reports pieces or entries. The Q&A does not give a verified count of unique riders represented by those items.




