Front Page

Supply Chain Forensics

OAuth Token Hijacking via Third-Party AI Tool Exposes Vercel's Internal Credential Vault

Infrastructure platform Vercel disclosed on April 18 that threat actors accessed internal systems by compromising Context.ai — a third-party AI assistant used by Vercel engineers. Attackers hijacked OAuth tokens tied to employee Google accounts, bypassing MFA controls entirely. Customer data including API keys and internal project metadata was extracted before the breach was contained and disclosed.

The attack reveals a critical forensic gap: OAuth delegation chains rarely surface in traditional SIEM logs. Investigators reconstructed the timeline across three separate log planes — Google Workspace audit events, Context.ai API access records, and Vercel's internal credential management system — none of which were correlated in standard IR playbooks. Third-party AI tooling now represents a privileged-access surface most organizations have not yet threat-modeled.

Critical Infrastructure

Apache ActiveMQ RCE Exploited at Scale — 6,000 Exposed Instances Targeted, CISA KEV Triggered

CISA added CVE-2026-34197 to its Known Exploited Vulnerabilities catalog this week as active exploitation of Apache ActiveMQ Classic reached critical scale — over 6,000 exposed instances identified via internet scanning with no patch applied. The flaw exploits improper input validation in the OpenWire protocol, enabling unauthenticated remote code execution with no user interaction required; CVSS 8.8.

ActiveMQ is embedded in enterprise middleware stacks as a message broker, making it largely invisible to endpoint security tooling. Forensic responders should prioritize ActiveMQ broker logs for unexpected subprocess trees spawning from the JVM process, unusual outbound connections from the broker host, and Java-based web shells staged in the ActiveMQ webapps directory. Post-exploitation persistence has favored scheduled tasks in observed cases.

Threat Bulletin

ACTIVE EXPLOIT
CVE-2026-32201

Microsoft SharePoint Server — authentication bypass enabling spoofing and unauthenticated RCE; patched in the April 14 Patch Tuesday emergency release and confirmed exploited in the wild prior to the patch. Forensic Note: Review SharePoint ULS logs for anomalous authentication events and check MSSQL audit trails for unauthorized stored procedure execution under SharePoint service accounts.

ACTIVE EXPLOIT
CVE-2026-34197

Apache ActiveMQ Classic — unauthenticated RCE via improper OpenWire protocol validation; CVSS 8.8. Added to CISA KEV April 2026 with over 6,000 exposed instances confirmed online. Forensic Note: Examine broker logs for unexpected JVM subprocess trees; scan webapps directories for Java-based web shell artifacts planted post-exploitation.

Malware Spotlight

BASANAI Ransomware (MedusaLocker)

Emerging MedusaLocker-family ransomware variant active April 2026, targeting enterprise Windows environments via RDP brute force and phishing. AES-256 / RSA-2048 encryption with .basanai extension; VSS deleted pre-encryption to block shadow copy recovery. Forensic Note: Hunt Event ID 4625 RDP brute-force sequences and vssadmin / wmic VSS deletion commands appearing in Windows event logs ahead of file encryption timestamps.