Cloud
Attacks against cloud identity and productivity platforms — Microsoft 365, Google Workspace, Azure, AWS — including OAuth consent phishing, device-code abuse, token theft, and the "identity-first" intrusion patterns that treat the tenant, not the endpoint, as the ground floor.
Elasticsearch Patches RCE in Machine Learning Module
Elastic patches CVE-2026-72649, a CVSS 8.8 deserialization flaw in the ML component that lets privileged users achieve RCE via crafted model artifacts.
Attackers Exploit Langflow to Steal OpenAI, AWS Keys
CVE-2026-0768, a critical unauthenticated RCE in Langflow, is being actively exploited to drain AI API keys and cloud credentials from exposed instances.
Nuclear Records Stolen via ownCloud Flaw, CISA KEV Added
CISA KEV added CVE-2023-49105 after nuclear data theft in Philippines. Self-hosted ownCloud users face an August 30 remediation deadline.
Next.js Patches Two Critical RCEs: Update Now
Vercel patched two critical unauthenticated RCE flaws in Next.js 15 and 16: one triggered by crafted AVIF images, another affecting Windows-hosted servers. No workaround exists for the Windows flaw — patch to 15.5.24 or 16.3.3.
StackGres CVSS 9.9 Bug Escalates DB Tenant to Admin
CVE-2026-78155 (CVSS 9.9): StackGres Kubernetes operator lets a low-privilege database tenant escalate to administrator. Patch immediately.
Fortune 500 Firms Named in Azure Data Theft Claim
Threat actor claims mass exfiltration from McDonald's, TCS, Vodafone, and other Fortune 500 firms via Azure. Named companies have not confirmed. Story developing.
Beacon CRM Breach Hits 1,000+ Charities via AWS Key
Over 1,000 UK charities had supporter data exposed after attackers used an AWS access key found in Beacon's public JavaScript build artifacts.
Azure CosmosEscape Flaw Exposed Any Tenant's Database
Wiz's CosmosEscape attack chain escaped Azure Cosmos DB's Gremlin sandbox, gained platform code execution, and extracted a key granting cross-tenant read/write access. Now patched.
NodeBB Patches Eight AI-Found High-Severity Flaws
Eight high-severity NodeBB flaws expose admin access and private chats in all pre-4.14.0 versions. Aikido Security's AI pentest found them in six hours. Patch to 4.14.2.
AI Agents Uncover Redis Zero-Days, Seven Patches Ship
Kimi K3 AI agents found authenticated RCE flaws in four Redis versions. Redis shipped seven security releases on July 23 — update your deployments.
FedRAMP 20x Ends Point-in-Time Authorization
FedRAMP 20x moves federal cloud authorization from periodic 3PAO assessments to continuous, machine-readable control evidence — what that shift requires from cloud operators.
NadMesh botnet raids exposed AI tools for 3,811 AWS keys
A Go botnet called NadMesh, active since early July, feeds a Shodan queue into ComfyUI, Ollama, n8n, Open WebUI, Langflow, and Gradio. Operator dashboard claims 3,811 AWS keys.
AWS persistence: four patterns to hunt after an incident
Rapid7's Jan Blažek maps four AWS persistence classes — new IAM users, assume-role edits, Lambda backdoors, federated tokens — with the CloudTrail signals to hunt for each.
Miggo: RabbitMQ leaked OAuth secret via obsolete endpoint
Miggo disclosed two RabbitMQ flaws today: an obsolete /api/auth endpoint exposed the broker's OAuth client secret, and a bug bypassed vhost boundaries.
CISA postmortem: nine alerts ignored, six months exposed
CISA's postmortem on its own six-month GitHub credential leak faults slow key rotation and nine ignored GitGuardian alerts — signal without intake.
WriteOut: One Preview Link Took Over Writer AI Accounts
SAND Security's WriteOut let a Writer AI agent preview link steal a signed-in user's session cookie across tenants. Writer has patched — the pattern hasn't.
GitLost: Public Issue Leaks Private GitHub Repo Data
Noma Security's GitLost shows how a public GitHub issue can trick Agentic Workflows into leaking private repos. Not patchable — scope your agent tokens today.
Dialogflow's Rogue Agent Flaw Is a Very Old Bug Class
Varonis' Rogue Agent finding in Google Dialogflow CX is a shared-runtime exec() escape — a bug class old enough to have graduated shared hosting.
Umbrij: ToddyCat Hijacks Gmail OAuth via Browser
Kaspersky Securelist detailed Umbrij, a ToddyCat post-compromise tool that self-grants Google Workspace OAuth tokens by driving a logged-in Chromium session. Nothing to patch. Plenty to audit.
ARToken PhaaS Targets M365 Device-Code Phishing
Cisco Talos exposed ARToken, a React-panel phishing-as-a-service tied to EvilTokens. Device code flow is the mechanic. Conditional Access is the fix, and most tenants still haven't turned it on.
ConsentFix + ClickFix: M365 Grants Outlive Resets
BleepingComputer covered two M365 hijack patterns and Opera's Paste Protect defense this week. The clipboard lane can be closed. The OAuth grant substrate underneath is unchanged.
Unpatched Argo CD Flaw Lets Unauth Cluster Takeover
Synacktiv disclosed an unpatched code-execution flaw in Argo CD's repo-server component. No fix, no CVE. Reachability of the internal port is the whole game.