AWS Compliance Gains, CISA Patch Orders, and AI Security Moves

Cloud providers expanded compliance coverage and disaster recovery capabilities, while security agencies issued urgent patch directives for actively exploited enterprise software. Vendors also shipped operational upgrades that reduce toil at scale, and researchers detailed evolving intrusion tradecraft alongside AI-specific risks and controls. The day’s developments collectively underscore a dual focus: hardening critical platforms and streamlining secure operations.

GovCloud Compliance and Backup Resilience

AWS update confirmed that Amazon Managed Grafana has achieved FedRAMP High authorization in AWS GovCloud (US-East and US-West), removing a key compliance barrier for public sector teams that need a managed visualization and alerting service for operational metrics across AWS and hybrid environments. In parallel, AWS update brought Amazon Aurora DSQL into scope for FedRAMP Moderate in US East (Ohio, N. Virginia) and US West (Oregon), enabling federal and regulated customers to adopt a serverless, distributed SQL database with active-active availability and multi-Region strong consistency under a standardized federal security posture.

AWS update extended AWS Backup restore testing to six additional Regions, allowing automated, periodic restore validation with scheduled tests, automated recovery point selection, and timing metrics against defined RTOs—useful for audit readiness and incident preparedness. Complementing that, AWS update expanded logically air-gapped vault support to six new Regions, adding options for immutable, isolated backups with default lock, encryption via AWS-owned or customer-managed keys, cross-account/Region copy, vault sharing through AWS RAM, and multi-party approval to help protect recovery assets even in account-compromise scenarios.

AWS Storage and Compute: Streamlining Operations

AWS update removed the 30-day minimum before transitioning objects from S3 Standard to S3 Standard-IA and S3 One Zone-IA, allowing immediate (0-day) lifecycle transitions. This change can lower storage costs for data that cools rapidly—such as backups, logs, or compliance archives—while maintaining millisecond access and aligning policies to actual access patterns.

AWS update added system-generated tags to Amazon S3 Event Notifications delivered to EventBridge, SQS, SNS, and Lambda. Teams can now use a single, tag-based EventBridge rule to filter across thousands of buckets without enumerating names, simplifying centralized event routing and reducing operational overhead.

AWS update for Amazon EC2 now surfaces the public Systems Manager Parameter Store parameters associated with public AMIs directly in AMI metadata. By returning the public SSM parameter via DescribeImages, administrators can reference stable parameters rather than hard-coding AMI IDs, streamlining image versioning and safer rollouts across automation and tooling.

AWS update introduced EC2 High Memory U7in-24TB instances in the Europe (Paris) Region, delivering 24 TiB of DDR5 memory and 896 vCPUs with high network and EBS bandwidth. Positioned for in-memory databases and analytics such as SAP HANA and major RDBMS platforms, AWS cites up to 45% better price performance versus prior-generation High Memory instances.

Exploitation, Patch Directives, and Legal Outcomes

BleepingComputer reported that CISA ordered federal agencies to remediate an actively exploited Oracle E-Business Suite flaw (CVE-2026-46817) by Saturday under Binding Operational Directive 26-04. The vulnerability in Oracle Payments’ File Transmission component allows unauthenticated takeover via low-complexity network attacks; Oracle patched the issue in its May 2026 Critical Patch Update.

CSO Online detailed CISA’s urgent advisory on on-premises SharePoint exploitation (CVE-2026-32201, CVE-2026-45659, CVE-2026-56164), urging immediate patching, AMSI integration, threat hunting for compromise, and rotation of machine keys when warranted. CISA added CVE-2026-56164 to its KEV catalog and emphasized that well-known “N-day” issues remain attractive targets.

CSO Online also covered Zoom patches for a critical account-takeover bug and three privilege-escalation flaws. The takeover issue initially impacted multiple Windows clients and VDI releases and could be exploited over the network without user interaction; Zoom identified and remediated the issues internally.

BleepingComputer reported that 23andMe will pay $18 million in a multistate settlement over a 2023 credential-stuffing breach that exposed genetic and ancestry data of 6.9 million customers. Investigators cited missing safeguards including MFA, password blocklisting, and adequate detection, and imposed strengthened controls, advisory oversight, and risk analysis protocols.

Infosecurity covered sentencing for two individuals who pled guilty to a cyber-attack on Transport for London, with the judge citing “selfish bravado.” Authorities linked broad disruption to the incident, including service shutdowns and password resets, and the NCA described the prosecution as the largest cybercrime case to reach UK courts.

Active Campaigns and AI Security

Fortinet analyzed a global campaign using obfuscated JScript droppers and AutoIt/LuaJIT loaders disguised as TrueType font files to deliver payloads such as Remcos, Agent Tesla, XWorm, Rescomms variants, and a Snake-derived keylogger. The multi-stage, fileless techniques emphasize anti-tampering, in-memory execution, and evasion of static and AI-based detection.

Microsoft blog documented increased ACR Stealer activity, including two intrusion chains initiated by ClickFix lures. The campaigns use WebDAV-hosted DLLs and fileless MSHTA-driven payloads, reflective in-memory execution, DPAPI browser credential harvesting, and document enumeration, with recommendations to monitor for WebDAV, MSHTA, obfuscated PowerShell, and in-memory behaviors.

The Hacker News highlighted ANY.RUN research on a PhantomEnigma campaign that hijacked over 20 Brazilian government websites and compromised email accounts to deliver modular backdoors via patched Electron apps. Trusted infrastructure and authenticated messages reduced suspicion, complicating detection and enabling credential theft, persistence, and potential fraud.

Palo Alto announced the general availability of Prisma AIRS AI Gateway, positioned as an enterprise AI control plane that unifies governance, identity, and runtime security for agentic AI interactions. Capabilities include observability, centralized governance for approved models/tools, scoped credentials, a Universal API for traffic routing, verifiable ephemeral agent identities, and inline inspection aligned with OWASP LLM and agentic application guidance.

Google Cloud published a GKE blueprint for securing AI workloads across infrastructure, supply chain, and application layers. Highlights include Confidential GKE Nodes, zero-trust networking, AI-focused bills of materials (k8s-aibom), Model Armor for prompt/response inspection, a GKE Inference Gateway for observability and quotas, and a phased adoption path from baseline controls to automated enterprise guardrails.

Infosecurity summarized Cato Networks research showing that a single high-level prompt to GPT-5.5 could drive an agent to execute a full attack chain in a controlled Active Directory lab, reaching admin-level privileges in about 40 minutes across tests. The findings do not unveil novel techniques but illustrate how frontier models can accelerate and scale known offensive workflows.

The Hacker News covered new research on Agent Data Injection (ADI), which corrupts trusted metadata fields to induce AI agents to misclick or run attacker commands without explicit instructions. Mitigations such as unguessable element IDs and full provenance tracking showed promise, though the latter carries usability costs; the researchers disclosed findings and released benchmarks to aid testing.