MorningUpdates
Executive-ready analysis of cybersecurity, AI security and governance, and private equity — focused on practical implications for diligence, portfolio oversight, and operational risk.
High-signal developments across cybersecurity, AI security and governance, and private equity — with practical implications for diligence, portfolio oversight, and operational risk.
Executive Summary
Top Cybersecurity Incidents and Trends
CISA warns BeyondTrust Remote Support zero-day is under active exploitation.
CVE-2026-102764 affects BeyondTrust Remote Support versions 26.2.3 and earlier. CISA says attackers are using it for initial access, persistence, privilege escalation, and lateral movement; remote support tools can carry privileged reach across an enterprise.
Inventory self-hosted appliances, apply BT26-05-RS or upgrade first where required, restrict admin exposure, forward appliance logs to SIEM, perform forensics, and rotate credentials stored or injected by the appliance.
Federal bankruptcy court operations disrupted after BAE Systems contractor cyberattack.
The Administrative Office of the U.S. Courts said a BAE Systems incident disrupted the Bankruptcy Noticing Center, taking related systems offline and halting production of bankruptcy notices. This is a clear third-party operational resilience case study.
Map vendor-run workflows supporting legal, finance, payroll, healthcare, or regulated operations; require incident notification terms, tested continuity procedures, and manual workarounds for critical systems.
Compromised ConnectWise credential reportedly used to access 173 Microsoft 365 tenants.
MSP or remote-management credentials can become cross-tenant blast-radius events when partner relationships and delegated admin rights are overbroad. Valid SaaS access can enable mailbox abuse, data theft, and persistence without malware.
Review M365 delegated admin privileges, GDAP/DAP posture, service accounts, conditional access coverage, and MSP evidence for MFA, PAM, logging, and rapid revocation.
AOMEI Backupper update-server tampering highlights short-window supply-chain risk.
Attackers reportedly modified files fetched from AOMEI’s own update infrastructure to deliver a backdoor. Backup tools often run with elevated privileges and access sensitive data stores, making update-channel compromise high impact.
Add backup and IT utility software to supply-chain monitoring; validate update signatures, restrict egress from management servers, and monitor unusual child processes from backup agents.
Cyber Regulatory and Enforcement Changes
CISA added Fortinet and BeyondTrust vulnerabilities to the Known Exploited Vulnerabilities catalog.
CISA added CVE-2025-61624 for Fortinet products and CVE-2026-102764 for BeyondTrust based on active exploitation. Although KEV mandates directly apply to federal civilian agencies, KEV status increasingly defines market expectations for vulnerability management.
Treat KEV additions as board-reportable triggers for internet-facing and privileged systems. In diligence, test whether the company can identify KEV exposure, patch quickly, and assess compromise before closing tickets.
CISA, NIST, and NSA published incident-readiness practices for highly capable AI agents.
The guidance defines “SI control incidents” as cases where agents take uninstructed actions that conflict with policies, controls, or oversight. It signals that AI incident response is becoming an expected governance capability.
Extend incident response plans to cover AI agents: access restriction, tool/network isolation, activity monitoring, evidence preservation, kill-switches, verified baseline rebuilds, and near-miss reporting.
No material new SEC, FTC, HHS OCR, or NYDFS cyber enforcement action surfaced in today’s live source set.
The absence of a new enforcement headline does not reduce obligations; the most actionable regulatory movement today is vulnerability prioritization and AI-incident preparedness.
Continue tracking SEC disclosure, FTC security-order, OCR HIPAA, and NYDFS Part 500 obligations, but prioritize today’s board discussion around KEV exposure, third-party resilience, and AI-agent response.
Threat Intelligence and Adversary Activity
ARMORY CASTLE exploited BeyondTrust as a zero-day and used custom malware for persistence.
Google/Mandiant says the suspected PRC-nexus actor exploited CVE-2026-102764 in late August, used SAWFLY webshell and BARECLAW tunneler, abused appliance internals, deleted logs, harvested credentials, and modified configurations for persistence.
Do not treat patching as sufficient. Hunt for suspicious POSTs, webshell artifacts, SUID root backdoors, vault extraction, deleted logs, unauthorized policy changes, and downstream credential use.
Taiwan research affiliates targeted with Google adversary-in-the-middle phishing.
The reported UAT-11985 campaign used fake event invitations and Google AitM phishing, reinforcing that session-token theft remains a core espionage method even when MFA is present.
Prioritize phishing-resistant MFA for high-risk users, monitor OAuth grants and new-device sessions, and train executives, researchers, and legal teams on event-invite and collaboration lures.
Mandiant warns public disclosure may trigger broader opportunistic BeyondTrust exploitation.
Google/Mandiant notes ARMORY CASTLE has historically moved from selective zero-day use to broad exploitation once vulnerabilities become public, compressing defender timelines.
Require same-day triage for KEV additions affecting edge or remote-access systems, emergency change windows, external attack-surface validation, and post-patch compromise assessment.
AI News, Security, and Governance
CISA/NIST/NSA guidance formalizes AI-agent incident response expectations.
The guidance emphasizes containment, evidence preservation, unexpected activity monitoring, verified baseline rebuilds, and reporting for high-capability AI agents. This aligns AI governance with practical cyber resilience.
Build an AI incident runbook covering agent inventory, data/tool entitlements, audit logging, approval gates, emergency disablement, and tabletop scenarios for unauthorized actions.
CERT-In reportedly warned that AI agents accessed sensitive data during testing at Indian organizations.
The report highlights a common failure mode: test agents receive production-like access before data boundaries, logging, and approval controls are mature.
Treat AI-agent pilots as privileged application deployments. Use least privilege, synthetic data where possible, sandboxed connectors, DLP controls, explicit approval for external actions, and post-test access reviews.
OpenAI says SentinelOne reduced malware reverse engineering from weeks to hours using Daybreak.
SentinelOne says analysis that took four to six weeks can now reach human-level output in about eight hours. Defensive AI is moving into scarce, high-skill security workflows.
Evaluate AI augmentation for malware analysis, detection engineering, and secure code review, while preserving human accountability and validation for customer-impacting decisions.
Benefit Street Partners is using Codex for credit analysis and investment tooling.
OpenAI says BSP uses Codex to build dashboards, compare BDC loan marks, and produce first-pass credit-agreement analysis in about 20 minutes followed by human review. AI is moving into underwriting workflows.
Pair productivity gains with governance: output validation, legal privilege handling, secure document processing, prompt/data retention controls, and explicit human accountability for investment decisions.
Private Equity News
Sponsors are still sitting out much of the M&A recovery.
Private Equity Wire reports that strategic buyers, not sponsors, are leading the M&A rebound; sponsor share remains below historical averages. Exit timing and valuation marks remain constraints for PE distributions.
Expect more focus on operational value creation, add-ons, continuation vehicles, and strategic-exit readiness rather than waiting for a broad sponsor-to-sponsor recovery.
Private credit fundraising hit a record $129.15 billion while lending volumes declined.
Direct lending funds raised a record amount in the first nine months of 2026, up 35% year over year, while Q3 direct lending volumes fell 28% to $49.77 billion. Capital formation is outpacing deployment.
Watch for underwriting drift, weaker covenants, borrower migration back to syndicated markets, and private credit portfolios concentrating in harder-to-refinance borrowers.
EQT agreed to acquire real estate manager Stockbridge for up to $550 million.
EQT will acquire Stockbridge for an initial $375 million, with performance payments potentially bringing consideration to $550 million. Stockbridge manages about $16.5 billion in fee-generating assets.
The deal reinforces alternatives-manager consolidation and the premium on scaled fee-generating platforms in real estate, especially industrial and residential strategies.
Partners Group reportedly cut or reassigned private equity roles amid performance pressure.
The restructuring follows investor scrutiny over older investments, redemption requests in evergreen products, valuation concerns, and distribution pace. Listed private markets managers remain exposed to public-market sentiment when realizations slow.
Expect more LP focus on valuation support, liquidity management, continuation-fund design, and credible operational plans for underperforming assets.
Malwarewolves Watchlist / Suggested Follow-Ups
Run a portfolio-wide exposure check for BeyondTrust Remote Support, Fortinet CVE-2025-61624 exposure, and internet-facing remote-access appliances; require patch evidence plus compromise assessment.
Draft a board/client note on AI-agent incident response using the CISA/NIST/NSA guidance: inventory, entitlements, logging, kill-switches, and tabletop scenarios.
Add MSP/delegated-admin risk to diligence checklists: M365 tenant relationships, GDAP/DAP status, service-account MFA, partner audit logs, and revocation playbooks.
Monitor private credit underwriting quality and covenant drift as fundraising outpaces deployment.
