Private Client · North America
Snowflake, Ticketmaster, and the Cost of Skipping MFA: A Cloud Security Post-Mortem
The same attack that compromised 560 million Ticketmaster records hit one of our clients. Here's the full post-mortem: what went wrong, how we contained it, and what we changed so it can't happen again.
Measured results
- Identified and contained the breach within 6 hours
- Enabled MFA across all Snowflake and cloud service accounts
- Reduced exposed credentials by 95% through vault rotation
- Passed SOC 2 audit 3 months later with zero findings
Challenge
The blocker
A SaaS company using Snowflake for analytics discovered their customer data was being sold on a dark web forum. They had no MFA enabled. The same attack pattern that hit Ticketmaster and Santander had hit them.
Solution
The approach
We conducted a full forensic investigation, contained the breach, enabled MFA and network policies on their Snowflake account, and rebuilt their access controls from scratch.
Execution story
In May 2024, news broke that Ticketmaster had suffered a data breach affecting up to 560 million customer records. Santander Bank confirmed a breach affecting 30 million customers. The common thread: both used Snowflake for cloud data storage, and both had accounts without multifactor authentication.
A threat actor group called UNC5537 (also linked to ShinyHunters) had used stolen credentials — purchased from infostealer malware — to access Snowflake customer accounts that had single-factor authentication. Snowflake, CrowdStrike, and Mandiant later confirmed: "This appears to be a targeted campaign directed at users with single-factor authentication."
One of our clients was on the list.
The call
The CTO called us on a Thursday morning. Their security team had found a listing on a dark web forum offering a dataset that matched their customer table. The listing included a sample of 100 records that were accurate. They used Snowflake. They didn't have MFA enabled on their service account.
We were on a plane within two hours.
What happened
The attack was simple, which is what made it effective:
-
Credential theft. An infostealer malware infection on a developer's laptop (months earlier) exfiltrated the developer's Snowflake credentials from their browser. The developer had left the company, but the credentials were still valid.
-
Single-factor authentication. The Snowflake account used username and password only. No MFA. No IP restrictions. No network policies.
-
Access. The attacker used the stolen credentials to log in to the Snowflake account directly. No exploit. No vulnerability. Just valid credentials and an account that didn't require a second factor.
-
Exfiltration. The attacker queried the customer table, exported the results, and listed the data for sale.
"We have not identified evidence suggesting this activity was caused by a vulnerability, misconfiguration, or breach of Snowflake's platform. This appears to be a targeted campaign directed at users with single-factor authentication." — Snowflake, CrowdStrike, and Mandiant joint statement
The whole attack took less than an hour from login to exfiltration. The credentials had been sitting in an attacker's database for months, waiting to be used.
Containment (hours 1-6)
The first six hours were about stopping the bleeding:
| Hour | Action |
|---|---|
| 1 | Rotated all Snowflake credentials. Disabled the compromised service account. |
| 2 | Enabled MFA on all remaining Snowflake accounts. |
| 3 | Added network policies restricting Snowflake access to known IP ranges. |
| 4 | Reviewed query history to identify exactly what was accessed. |
| 5 | Notified legal counsel and began regulatory notification assessment. |
| 6 | Confirmed the attacker's session was terminated and no persistence was established. |
The key finding from the query history review: the attacker had accessed one table — the customer table — and nothing else. They were after data to sell, not system access. This narrowed the scope of the breach significantly.
Root cause and remediation (days 1-30)
The root cause was straightforward: a valid credential with no second factor. The fix was equally straightforward, but the implementation touched every part of their cloud security posture.
MFA everywhere
We enabled MFA on every cloud service account: Snowflake, AWS, GitHub, their CI/CD pipeline, their monitoring tools. Not just admin accounts. Every account, including service accounts where possible.
For service accounts that couldn't use interactive MFA, we moved to short-lived credentials managed by a secrets vault. No more long-lived passwords sitting in CI configuration files.
Network policies
We added network policies to the Snowflake account restricting access to the company's office IP ranges and VPN. Access from any other IP is blocked by default and requires explicit allowlisting.
Credential rotation
We rotated every credential that had been in use for more than 90 days. The compromised credential had been valid for 7 months because nobody had thought to rotate it when the developer left. We put a 90-day rotation policy in place with automated reminders.
Offboarding process
The deeper issue was that the developer's credentials were still valid 7 months after they left. We rebuilt the offboarding checklist to include credential revocation for every service the developer had access to, with a signed-off confirmation from IT.
The broader pattern
Our client wasn't the only one. The Snowflake breaches hit at least 165 companies, including AT&T, Advance Auto Parts, and Neiman Marcus. The multidistrict litigation consolidated in Montana covers over 500 million affected individuals.
The pattern in every case was the same:
- Stolen credentials from infostealer malware
- Snowflake accounts with single-factor authentication
- No network policies restricting access
- Data exfiltration and sale on dark web forums
"Neither Snowflake nor Ticketmaster implemented three of the most basic cybersecurity policies to protect personal data, including most prominently, multifactor authentication." — Court filing, Snowflake Data Security Breach Litigation
The numbers
| Metric | Before | After |
|---|---|---|
| MFA coverage | 0% of cloud accounts | 100% of cloud accounts |
| Network policies | None | IP-restricted on all data stores |
| Credential rotation | Never (7-month-old creds in use) | 90-day automated rotation |
| Time to detect breach | Unknown (discovered via dark web) | < 1 hour (anomaly detection) |
| SOC 2 findings | N/A | 0 findings, 3 months post-incident |
What we learned
This breach wasn't sophisticated. It didn't use a zero-day. It didn't exploit a vulnerability. It used stolen credentials on an account that didn't require a second factor. That's it.
The fix is not complex. MFA on every account. Network policies on every data store. Credential rotation. Proper offboarding. These are basics. The Snowflake breaches happened because companies skipped the basics.
If you're using Snowflake, AWS, or any cloud service and you don't have MFA enabled on every account, enable it today. Not next week. Today. The credentials to your account may already be in an attacker's database, waiting to be used. The only thing standing between you and a breach is a second factor.