Back to Exposure Report
590.69 GB claimed, 108,426 files October 22, 2026 • Professional Services

PwC

The data was not where the firm's defences were, because it was not supposed to be anywhere important.

Employee recordsEmployment and training informationConfidential audit documentsSingle sign-on source code
1

What happened?

The Everest ransomware group listed PwC as a victim, with the listing reported on 22 October 2026, claiming 590.69 GB of data. A public alert described 108,426 files allegedly taken from internal HR and learning systems, with the alleged data spanning PwC operations in Australia, China, Canada and the United Kingdom.

PwC publicly acknowledged an incident on 23 October 2026, stating it affected a third party vendor environment used to host employee learning records, and that the affected site does not house client data, sensitive personal information or financial records. The group's broader claims remain unverified.

2

What data was actually inside?

According to the group's claim: employee records, employment and training information, confidential audit documents, and single sign on source code. One HR dataset is said to contain 213,084 records. These figures come from the attacker and have not been independently confirmed.

PwC's own characterisation is narrower and describes employee learning records in a vendor environment. The gap between the two descriptions is the story, and it will take weeks to close.

If the single sign on source code claim holds, that item is categorically different from the rest. Source code for an authentication path is not a record about a person. It is a description of how an identity system decides who to trust, and its value to an attacker outlives any individual file in the set.

3

Who gets hurt and how?

Employees and former employees across four countries, whose training and employment records sit in a learning platform they were required to use.

Learning records look harmless and are not. A completion history describes what a person was remediated on, which courses were assigned after an incident, and which compliance modules were mandated. In a professional services firm those assignments can imply a disciplinary context. Employment information adds the identifiers that make the rest usable.

If audit documents are genuinely in the set, the exposure reaches past PwC's own staff to the organisations those audits concern, none of whom were party to the vendor arrangement.

4

What did they think they were doing right?

Separating the learning platform from client systems. PwC's statement emphasises exactly this, and the emphasis is appropriate, because for an audit firm the catastrophic scenario is client data and this does not appear to be that.

The decision that created the exposure was a sound one. Employee training is a commodity function, commodity functions belong with specialist vendors, and a learning management system is about as low risk as a corporate application gets. The firm put its lowest sensitivity workload in its least defended environment, which is what a rational risk ranking tells you to do.

5

What did they not know about their own data?

What had accumulated in the low sensitivity environment. A learning platform starts as course completions and then quietly becomes an HR system, because the course assignments need employee records, and the employee records need identifiers, and the reporting needs organisational structure. Four countries of that accumulation is a dataset nobody specified.

The alleged presence of confidential audit documents and single sign on source code in an employee learning environment, if confirmed, would be the sharpest version of the point. Neither belongs there. Neither was put there by a decision anybody would defend. They arrive through the ordinary processes by which files end up adjacent to other files.

The vendor tenant is the structural weakness. It is inside the supply chain and outside the security programme, which means it inherits neither the firm's monitoring nor the firm's data classification. The TeamUpdraft entry in this series shows the same blind spot one tier further down.

If you use cloud storage, do you know what sensitive data lives in your buckets and blobs? Or would you find out the same way they did?

Not sure where to start? How to find sensitive data across your estate

6

What does attribution look like the morning after?

Everest has made the claim. PwC has acknowledged a vendor incident with a narrower description. Both can be accurate at once, and resolving the difference requires the firm to inventory a tenant it does not operate.

That is the hard part. Notification obligations across Australia, China, Canada and the United Kingdom diverge on what counts as personal data, what counts as an eligible breach, and how fast the clock runs. Each answer depends on the field list, and the field list lives in a vendor environment. The claimed deadline attached to the listing is attacker generated and should not be treated as a timetable.

7

What would have changed the outcome?

An inventory of what the learning platform had actually collected, as opposed to what it was commissioned to hold. The commissioning decision was correct. The drift afterwards was never reviewed.

Second, treating the vendor tenant as in scope for data discovery. A system classified as low sensitivity at procurement is not low sensitivity four years and four jurisdictions later, and nothing re-examines that classification unless somebody schedules it.

PwC found out the hard way.

Your team could spend the next 6 months rebuilding systems, notifying customers, and answering legal questions. Or you could spend 24 hours finding out what's actually at risk.

Want to understand the process first? Read our guide to sensitive data discovery .