Snowflake Cloud Data-Theft and Extortion Campaign (UNC5537)

Technical Analysis

The threat actors did not exploit a vulnerability in Snowflake itself; instead, they leveraged credentials obtained from prior infostealer malware infections and credential-stuffing campaigns, targeting accounts that lacked multi-factor authentication and had no IP allowlisting. Once authenticated, attackers used legitimate Snowflake query tools and automated scripts to bulk-export customer data, then demanded ransom payments to prevent public release or sale of the data. This case underscores the risks of SaaS/cloud data warehouse platforms as high-value targets when tenant-side security hygiene (MFA, credential rotation, network restrictions) is weak, rather than a flaw in the platform provider. Organizations running AI agents or RAG pipelines that source data from or write logs/embeddings into Snowflake-hosted warehouses could have sensitive training data, customer records, or API keys exposed if agent service accounts share the same weak authentication posture, making credential hygiene for any cloud data warehouse integrated into agent workflows a critical concern.

Affected Systems

Snowflake cloud data warehouse customer accounts lacking MFA and network access controls; at least 165 downstream organizations across various sectors using Snowflake for data storage and analytics

Indicators of Compromise

• N/A - no specific hashes, IPs, or malware file names disclosed in source reporting; associated with prior infostealer-derived credential dumps

Remediation Steps

1. 1

   Enforce MFA on all cloud data platform accounts

   Require multi-factor authentication for all Snowflake and similar SaaS data warehouse accounts, including service and agent-linked accounts.

2. 2

   Implement network policies/IP allowlisting

   Restrict Snowflake account access to known corporate IP ranges or VPN endpoints to reduce exposure to stolen credential reuse.

3. 3

   Rotate and audit credentials

   Regularly rotate API keys, service account credentials, and passwords used by human and automated (agent) systems accessing cloud data platforms.

4. 4

   Monitor for anomalous data exfiltration

   Deploy detection for unusual bulk query or export activity from data warehouse accounts, including AI agent or automation service accounts.

5. 5

   Audit third-party and agent integrations

   Review which AI agents, RAG pipelines, or automated tools have credentials or API access to cloud data warehouses and apply least-privilege principles.