Blind Spots: Two Cloud Threats You Didn’t Even Know You Had
summary
Research finds massive, concrete blind spots: a red team spins up an external account, enumerates public GitHub and Google artifacts, and with a misconfigured role trust policy grabs a temporary token and reads internal S3 buckets including encrypted credentials it then decrypts via KMS access demonstrating identity gives broader access than a host exploit. Large-scale scraping of ~300k files and 150k repos yields 70k role names and 33k accounts and reveals 42% of CloudFormation templates insecure and 60% of cloud storage has logging disabled. The talk cites an open source scanner IAMFinder and prescribes multi‑cloud visibility, auto‑remediation, shift left and runtime controls.
AI-written summary · about
chapters (13)
- 0:00Cloud security blind spots
- 1:26Cloud clarity areas
- 2:50Cloud IAM complexity
- 5:41Red team exercise results
- 11:47IAM misconfigurations globally
- 16:21Infrastructure as code security
- 17:49Insecure cloud formation findings
- 21:23Cloud formation misconfigurations
- 23:50Terraform security risks
- 26:39Improve cloud security
- 30:14Cloud security maturity model
- 33:42Cloud security optimization
- 35:57Conclusion
related talks
47mBreaking the Cloud to Rebuild It: A Tale of 3 ☁️ Breaches!
36mCloud Hacking Scenarios
33mCloud Hacking Scenarios
45mProtecting you from you: Misconfiguration-caused breaches
36mIAM Confused: Decoding 8 Real World Cloud Identity Breaches
46mStories from the cutting edge: Cloud security in 2023
57mCloud’s Dirty Little Secret: It Was Misconfigs All Along
19mThe Blind Spots of IT Experts on Security & What Companies Must Do Against It