DEV Community

Shubham Chaudhary
Shubham Chaudhary

Posted on

Trezor Data Breach: How a Metabase SQL Injection Exposed 80,000 Customers

 ## The Trezor Breach That Grew 6x Because of a Broken Vendor Promise

If you follow supply-chain and vendor risk incidents, this one is worth a technical breakdown — not just a headline skim.

TL;DR

On September 4, 2026, Trezor confirmed that a breach at fulfillment partner ShipMonk now affects 80,000+ US customers — up from the ~13,689 originally disclosed on August 13. The root cause: a critical SQL injection zero-day in Metabase, the BI/analytics tool ShipMonk used to query customer order data.

Trezor's hardware wallets, seed phrases, and private keys were not compromised. This was a contact-data exposure via a third-party vendor's analytics layer — not a cryptographic failure.

Attack Chain (MITRE ATT&CK mapped)

  • Initial Access — T1190, Exploit Public-Facing Application (Metabase SQLi)
  • Privilege Escalation — T1068, reaching admin-level access on the Metabase instance
  • Collection/Exfiltration — direct DB querying through the compromised BI layer

Other companies (Framework, Tally) disclosed similar Metabase-linked breaches around the same window — this looks like a broader campaign, not a Trezor/ShipMonk-specific attack.

The part that actually matters for engineers and SOC teams

Trezor's original disclosure was scoped around a 90-day data retention policy. Trezor said it had repeatedly requested — and received — written confirmation from ShipMonk that data older than 90 days had been deleted.

It hadn't been. Records dating back to November 2019 — nearly 5 years past expiration — were still live in ShipMonk's systems when the Metabase compromise happened.

This is the real lesson: a signed attestation is not equivalent to an audited deletion log. If your team manages vendor risk, "get it in writing" doesn't scale as a control — you need independent verification (audit logs, record counts, third-party confirmation).

If you run Metabase internally or through a vendor

  • Confirm you're patched against the Aug 2026 Metabase SQLi zero-day, and that sessions were invalidated, not just blocked going forward
  • Audit admin-level access logs on BI/analytics tools sitting in front of PII — these are chronically under-monitored vs. production DBs
  • If a vendor breach mentions SQL injection as entry vector, treat your BI layer as in-scope for investigation too

Full technical writeup with the complete timeline, exposed data breakdown, and prevention controls:
🔗 https://www.xpert4cyber.com/2026/09/trezor-shipmonk-data-breach-67000-exposed.html

What's your team's actual process for verifying vendor data deletion — not just requesting it? Curious how others are handling this.

Top comments (0)