Incident sheet · /security
- System
- Security & Risk
- Filed
- Read
- 3 min · 569 words
- Severity
- Routine
Attackers Deleted 100+ Storage Accounts in 7 Minutes — Cloud Resilience Tools Didn't Help
Microsoft logged a 7-minute attack deleting 100+ Azure storage accounts. Six cloud data protection myths — from S3 versioning to 'stateless' apps — explain why recovery plans keep failing.

Event log
- Microsoft's September 2026 Storm-3168 report: attackers attempted 100+ storage account deletions in ~7 minutes via compromised Azure identities
- January 2025: Codefinger ransomware used compromised AWS credentials and SSE-C encryption against S3 buckets
- Code Spaces shut down within days of a 2014 AWS console breach that wiped data, backups and snapshots
- Six identified myths: S3 versioning, multi-region replication, snapshots, IaC rebuilds, statelessness, data ephemerality
In a seven-minute destructive sequence detailed by Microsoft in September 2026, attackers holding compromised Azure identities attempted more than 100 deletions of storage accounts — and deleted most of them. The incident, tied to the Storm-3168 agentic-driven cloud attacks, also targeted recovery protections. It's the clearest evidence yet that the tools cloud teams trust to survive failure are not the tools that survive an attacker.
Cloud teams spend heavily on replication, versioning, snapshots, and infrastructure as code to keep applications available. The trouble starts when they assume those same capabilities allow recovery from a cyberattack. In conversations with customers, the pattern repeats: some pour money into disaster recovery for the next hyperscaler outage, while others assume data sitting in AWS or Azure is already protected. Both misunderstand what cloud-native resilience tools were designed to do — and those tools behave differently when the disruption is malicious.
What are the six myths — and what do they cost?
1. Amazon S3 versioning equals backup. Versioning helps when someone accidentally overwrites an object, not when an attacker holds your credentials. In January 2025, Halcyon researchers documented the Codefinger ransomware campaign: attackers used compromised AWS credentials and S3 server-side encryption with customer-provided keys (SSE-C) to make data inaccessible, then threatened deletion through lifecycle policies. Fix: restrict who can alter lifecycle policies, and use Object Lock or MFA Delete.
2. Multi-region replication protects you. A copy in another region isn't automatically a clean copy. Whether it's S3 replication or Azure Storage georedundancy, corrupted or encrypted data may replicate right alongside legitimate changes, and deletion behavior varies by service. Keep recovery copies in a separate account production credentials can't reach.
3. Snapshots are good enough. Code Spaces, a code-hosting service, learned this in 2014 when an attacker with its AWS control panel access deleted data, backups, machine configurations and snapshots. The company shut down within days. Lock recovery snapshots so no one — not even an admin — can delete them before retention expires.
4. IaC means we can rebuild anything. Terraform can rebuild database infrastructure, but restoring customer records and transaction history requires a separate data recovery solution. Recovery tests should restore the data the app depends on, then verify the restored environment actually works.
5. We're stateless, so there's nothing to protect. "Stateless" apps still write data in several places: Postgres, Amazon S3, DynamoDB, Redis, Snowflake, Apache Kafka. Map every service each application writes to, decide what must be recoverable, and assign ownership — otherwise data falls through the cracks between teams.
6. Our data is ephemeral, so losing it doesn't matter. Audit logs look expendable until an incident, when security teams need them to establish what the attacker touched and whether anything is safe to restore. Ask what losing a given dataset would cost — stalled investigations, interrupted operations, revenue impact, regulatory exposure — and put logs under retention policies with access controls that block compromised production credentials.
What does this change for cloud teams?
Two questions cut through all six myths: could someone with stolen production credentials delete your recovery copies, and have you ever actually restored from them?
In a cloud incident, the first executive question is "When will the business be back online?" — often while the response team still doesn't know how much is damaged. Know what your cloud tools can and can't recover before you get asked.
via infoworld.com/feed/ (Source)
More from Olivia Hart
Show full bio
News editor covering money, funding and market shifts around enterprise information technology and the leaders who run at CIO Monitor.
207 articles
The newsroom digest
One email a day with the stories that matter. No noise.
Cross-references
- 24 Hours Down, Zero Defections: Why Microsoft Outages Don't Cost It Customers
- Microsoft 365 Back After Multi-Day Outage Hit Firms in 70+ Countries
- Microsoft's Global Outage: A $7 Billion Lesson in Cloud Concentration Risk
- ConnectWise Suspends Cloud Services Following Cyberattack
- Microsoft 365 outage reportedly affects 100,000 users worldwide