Cloud-Security on CuraSec

9,300+ Leaked AWS Keys Still Active, Grant Full Account Control

Sat, 22 Aug 2026 11:32:44 +0000

Engineer — Act: Still-valid exposed AWS keys require no exploitation sophistication — the credential is the exploit. Audit all active IAM access keys in your AWS accounts, cross-reference against the leaked dataset, rotate any keys created or last-used anomalously, and enforce least-privilege policies with automatic key rotation going forward.
SOC/IR — Act: Active leaked credentials mean unauthorized access may already be occurring. Hunt CloudTrail logs since August 2022 for API calls from unexpected source IPs, new IAM user/role creation, or unusual resource provisioning that could indicate keys were already abused by third parties.
Leader — Act: Hundreds of corporate AWS keys with full-account-control scope being publicly available for up to four years is a material risk requiring same-week action — confirm whether your organization’s keys appear in the exposed set and direct engineering to complete a credential audit and rotation before end of week.

SANS ISC: Observed Scans Targeting Cloud Metadata Service (IMDS)

Thu, 20 Aug 2026 11:39:11 +0000

Engineer — Plan: Reinforces the need to enforce IMDSv2 (hop-limit 1, require session tokens) on all EC2/GCP/Azure VMs and audit IAM role assignments to minimize credential scope accessible via the metadata endpoint.
SOC/IR — Learn: No new IOCs or campaign detail here, but a useful reminder to verify detections exist for unusual internal requests to 169.254.169.254, which can indicate SSRF or compromised workload attempts to harvest credentials.
Leader — Skip

Snowflake Extortion Actor Pleads Guilty; 165 Orgs, 100M Records Affected

Fri, 07 Aug 2026 00:21:58 +0000

Engineer — Learn: The 2024 Snowflake credential-stuffing campaign is legally concluded with no new technical disclosures; reinforces that MFA enforcement on cloud data warehouses is non-negotiable, but no immediate action is required if controls were hardened after the original incident.
SOC/IR — Learn: The guilty plea closes attribution on a major 2024 campaign but surfaces no new IOCs, TTPs, or detection opportunities; useful for building institutional knowledge about the attacker’s methods (credential reuse at scale against SaaS platforms).
Leader — Learn: A high-profile case closure illustrating the scale of SaaS vendor risk when MFA is absent; valuable reference for board-level narratives on third-party cloud risk and regulatory exposure tied to customer data held by a vendor.

UNC6671 Vishing-AiTM Campaign Targets Financial Services, Enterprise Cloud

Fri, 07 Aug 2026 00:21:58 +0000

Engineer — Plan: Active group uses AiTM to bypass MFA on M365 and Okta; implement phishing-resistant FIDO2/hardware-key MFA and tighten Conditional Access or Okta device-trust policies to invalidate intercepted session tokens.
SOC/IR — Act: Active campaign with mappable TTPs — hunt for anomalous Okta and M365 session activity (unexpected token origins, bulk SharePoint/OneDrive exfil) since May 2026 and pull the GTIG report for infrastructure IOCs tied to Redact, Pink, Helix, and Falcon brands.
Leader — Act: Extortion group is actively hitting financial services, private equity, and professional services — if your org falls in these verticals, brief leadership this week on the campaign and verify that helpdesk impersonation and personal-device contact scenarios are covered in your security awareness program.

AWS, Google, and Vercel patch agent flaws that bypass model guardrails

Thu, 06 Aug 2026 13:03:19 +0000

Engineer — Plan: If you operate AI agents on AWS, Google, or Vercel infrastructure, audit your agent configurations and apply vendor patches; the core risk is that tool invocations can be triggered without a model turn, defeating system-prompt and content-filter controls you may rely on for safety.
SOC/IR — Learn: No IOCs or active exploitation reported, but this class of agent-layer authorization bypass is worth understanding as AI agent deployments grow — future detections may need to monitor tool-call events that lack a preceding model-turn record.
Leader — Plan: If your organization uses AI agent frameworks on these three platforms, confirm engineering teams have reviewed and applied patches; this also signals the need for an AI agent security policy that doesn’t assume model-layer guardrails are the last line of defense.

Canadian pleads guilty to Snowflake data-theft scheme hitting 165 orgs

Thu, 06 Aug 2026 13:03:19 +0000

Engineer — Learn: The guilty plea closes the legal chapter on a credential-stuffing campaign that bypassed MFA-less Snowflake accounts; no new vulnerability or patch, but reinforces ensuring MFA and session token controls are enforced on all cloud data warehouse accounts.
SOC/IR — Skip
Leader — Learn: The case confirms 165 organizations were breached through stolen credentials at a single cloud provider, a useful data point for board-level discussions on cloud vendor risk and MFA mandates — no immediate action required given the incident predates this plea.

Snowflake Hacker Pleads Guilty; 165 Orgs and 100M Records Exposed

Thu, 06 Aug 2026 13:03:19 +0000

Engineer — Skip
SOC/IR — Learn: The guilty plea closes the loop on a major credential-based cloud breach campaign; review whether your org’s Snowflake tenant MFA and network policies would have detected or blocked the access patterns used in 2024.
Leader — Learn: A high-profile conviction in a breach affecting 100M people and 165 orgs is useful context for board discussions on cloud vendor risk and credential-based attack exposure; no immediate action required unless your org was among those affected.

FedRAMP 20x Replaces Rev5 With Continuous Evidence Requirements

Fri, 24 Jul 2026 12:43:46 +0000

Engineer — Plan: If your team supports a FedRAMP-authorized product, start mapping how you’ll generate continuous, machine-readable control evidence — point-in-time assessment artifacts will no longer suffice once the transition deadline arrives.
SOC/IR — Skip
Leader — Plan: If your organization holds or pursues FedRAMP authorization, place the Rev5-to-20x transition on the roadmap this quarter: budget for tooling that produces continuous evidence and assess your current ATO timeline against the sunset date. Note this piece appears to be vendor-authored content from Anecdotes, so verify transition specifics against GSA primary sources.

Bit2Watt: Cloud GPU Tenants Could Destabilize Data Center Power Grids

Tue, 21 Jul 2026 12:43:35 +0000

Engineer — Learn: Novel academic research showing that ordinary tenant GPU workloads can modulate data center power draw enough to stress the upstream grid — no exploit or patch surface exists, but it reshapes how multi-tenant GPU infrastructure risk should be assessed in cloud architecture reviews.
SOC/IR — Learn: No IOCs, no active exploitation, and no practical detection surface for workload-level power manipulation; file as background awareness on an emerging side-channel class with no near-term hunt or rule-writing opportunity.
Leader — Learn: Early-stage academic research with no current exploitation; worth tracking as a long-horizon risk narrative around cloud infrastructure resilience and power-grid dependencies, but no board or customer communication is warranted now.

Mandiant: Hardening Publicly Exposed Serverless Cloud Functions

Thu, 16 Jul 2026 12:18:39 +0000

Engineer — Plan: Mandiant assessments routinely find unauthenticated Cloud Run/Functions exposed to the internet; audit your serverless inventory for missing auth controls and apply the hardening patterns (least-privilege service accounts, input validation, network egress restrictions) this quarter. No active exploitation signals elevate this to Act.
SOC/IR — Learn: The LFI/RFI and command-injection paths described could inform detection logic for serverless workloads, but there are no IOCs, no named campaign, and no novel TTPs here — no immediate hunt or rule-writing required.
Leader — Skip