On-Prem SharePoint Is Under Active Attack — And Patching Alone Will Not Remediate It (2026)
CISA confirmed active exploitation of four on-premises SharePoint Server flaws that let attackers steal IIS machine keys and plant persistence that survives patching. Here is the remediation order that actually evicts them, and the migration path off on-prem for Houston and The Woodlands firms.
Introduction
If your organization still runs a SharePoint Server in a closet, a colo cage, or a virtual machine somebody spun up in 2018, this is the post to read all the way through. The on-prem SharePoint Server exploit 2026 activity is not a theoretical patch-Tuesday item. It is an active intrusion campaign, and the single most important detail is one that most coverage buries: applying the patch does not remove the attacker.
Why patching alone will not remediate it
Here is the mechanic that makes this campaign different from an ordinary RCE. Once attackers get code execution on a SharePoint front end, the documented post-exploitation step is to steal the server's IIS machine keys - the ASP.NET validation and decryption keys that SharePoint uses to sign and encrypt ViewState and other serialized payloads.
With those keys in hand, an attacker no longer needs the vulnerability. They can forge a validly signed serialized object and send it to your server as a normal, well-formed web request. The server verifies the signature, deserializes the payload, and executes it. You patched the bug. The key you never rotated is now the backdoor. This is the same pattern security teams learned the hard way during the 2025 ToolShell campaign (CVE-2025-53770 and CVE-2025-53771), where attackers extracted the ValidationKey and DecryptionKey to forge __VIEWSTATE payloads that survived patching. The industry is repeating the mistake.
SharePoint IIS machine key rotation is the step everyone skips
SharePoint IIS machine key rotation is not optional cleanup. It is the step that actually closes the door. In SharePoint you rotate with the Set-SPMachineKey cmdlet per web application, use Update-SPMachineKey to deploy the keys to the other servers in the farm, or trigger the Machine Key Rotation timer job from Central Administration (Monitoring, then Review job definitions) - then restart IIS farm-wide with iisreset.exe so the worker processes pick up the new keys.
Worth knowing: automatic machine key rotation is available starting with SharePoint Server Subscription Edition version 25H1 and the September 2025 public update for SharePoint Server 2016 and 2019. If you are on those builds, confirm whether it is actually enabled rather than assuming it.
One ordering caveat that CISA calls out explicitly, and that DIY cleanups get backwards: hunt for and remediate intrusion artifacts, including machine-key harvesters, before rotating keys. If an attacker still has an active foothold when you rotate, they simply harvest the new key and you have accomplished nothing but a maintenance window.
The remediation sequence, in the order that works
- 1. Contain first. Get the server off the public internet today. CISA recommends avoiding direct internet exposure unless necessary, blocking external access to SharePoint Central Administration, restricting farm and database communication to required systems, and placing any necessarily exposed server behind a Layer 7 reverse proxy or similar application-layer control. Better still: put it behind identity-aware access so only authenticated, device-verified sessions reach it. Our take on that model is here: zero trust network access.
- 2. Patch and verify. You likely need more than one update: CVE-2026-45659 was fixed in the late-May 2026 out-of-band release, while CVE-2026-56164 and CVE-2026-58644 were fixed in the July 14, 2026 Patch Tuesday. Apply the correct updates for your exact build, then confirm the build number actually changed. Verification failures are common on multi-server farms, and CISA specifically calls out verifying successful installation.
- 3. Enable AMSI and antivirus. Turn on Windows Antimalware Scan Interface integration for SharePoint web applications - Microsoft's Full Mode scans request bodies, not just query strings - and run Microsoft Defender Antivirus (or an enterprise equivalent) on every farm node.
- 4. Hunt and remove artifacts. See the next section.
- 5. Rotate machine keys, then iisreset. Per web application, replicated across the farm.
- 6. Reset service account credentials. Farm account, application pool identities, search service, and any SQL account reachable from the SharePoint tier. If the box was owned, those secrets were in memory. This one is standard incident-response practice rather than a specific CISA instruction, but skip it at your peril.
- 7. Re-hunt to validate. Confirm no persistence survived the rotation. This is the step that proves eviction rather than assuming it.
SharePoint webshell persistence hunting: what to look for
SharePoint webshell persistence hunting means going beyond antivirus alerts. Concretely, look for unexpected .aspx or .ashx files in the LAYOUTS directory and other IIS-served paths with recent write times; IIS worker processes (w3wp.exe) spawning cmd.exe or powershell.exe; new or altered scheduled tasks running under SharePoint service identities; application pool identities that were changed or added; new local accounts and group memberships; and outbound connections from the SharePoint tier to hosts nobody can explain. CISA also advises establishing tailored logging to detect exploitation activity and reviewing telemetry for anomalous requests, suspicious worker-process activity, webshells, and machine-key access. Then check whether your logging retention even goes back far enough to answer the question. For most small and mid-size firms it does not, which is exactly the gap managed detection and response is built to close with continuous automated monitoring.
Forget the KEV four-day remediation deadline: the clock is now 72 hours
Federal civilian agencies operate under Binding Operational Directive deadlines when a CVE lands in the Known Exploited Vulnerabilities catalog, and those deadlines got shorter this year. BOD 26-04, "Prioritizing Security Updates Based on Risk," superseded and revoked the long-standing BOD 22-01 on June 10, 2026. For the highest-risk class of vulnerability - internet-exposed, KEV-listed, automatable - agencies now get roughly 72 hours to remediate and run a forensic triage to determine whether they have already been breached.
You can see it in this cluster's due dates, each three days from listing: CVE-2026-45659 added July 1 with a July 4 due date, CVE-2026-56164 added July 14 with a July 17 due date, CVE-2026-58644 added July 16 with a July 19 due date. Private companies are not legally bound by any of it. You should still treat 72 hours as the standard, because attackers are moving well inside that window and because auditors under SOC 2 and similar frameworks increasingly ask how fast you close known-exploited findings.
The Houston and Woodlands version of this problem
Across Greater Houston and The Woodlands, we see the same profile repeatedly: an engineering firm, a mid-size law practice, an energy-services company, or an architecture studio with a SharePoint or WSS document library that was stood up years ago for drawings, case files, or project records. Nobody owns it. It was never decommissioned because "somebody might still need the old CAD folder." And because it was set up before anyone thought hard about remote access, it is quietly published to the internet on port 443 with a certificate that still renews.
That is the exact target profile. Texas energy-services and engineering firms carry an additional wrinkle: those legacy libraries often hold client-confidential drawings and contract documents that trigger notification obligations and prime-contractor flow-down requirements if exposed, including NIST 800-171 obligations for anyone in a defense or federal supply chain.
When to call incident response instead of cleaning it up yourself
Stop and bring in incident response if you find any of the following: an unexplained .aspx file in a SharePoint-served directory, evidence of w3wp.exe launching a shell, a service account with a password you did not change, or logs that simply do not go back far enough to rule out compromise. Also escalate if the server was domain-joined with a privileged farm account, because SharePoint compromise has a well-documented path to broader domain compromise.
Know what that costs before you need it. Published market rates for digital forensics and incident response generally run about $175 to $600 per hour when you have a retainer in place, and roughly $500 to $1,500 per hour for emergency engagements without one. Annual retainers themselves commonly run $10,000 to $100,000 depending on prepaid hours and response SLA. That is real money. It is far less than the cost of a re-entry three months after you thought you were done, and dramatically less than a ransomware recovery engagement after the same access is sold on to a ransomware affiliate - which, given the Warlock activity already tied to this bug class, is the realistic end state.
Frequently Asked Questions
Is SharePoint Online affected by the 2026 SharePoint exploits?
No. SharePoint Online, the Microsoft 365 cloud service, is not affected by CVE-2026-32201, CVE-2026-45659, CVE-2026-56164, or CVE-2026-58644. These vulnerabilities affect on-premises SharePoint Server 2016, 2019, and Subscription Edition only.
If I already patched, am I safe?
Not necessarily. If an attacker reached your server before the patch and stole the IIS machine keys, they can continue forging signed serialized payloads that the patched server will still accept and execute. Patching closes the entry point but does not revoke stolen keys. You must hunt for artifacts, rotate machine keys, restart IIS across the farm, and reset service account credentials.
What does rotating IIS machine keys actually do?
It replaces the ASP.NET validation and decryption keys used to sign and encrypt serialized data such as ViewState. Any payload signed with the old stolen key is rejected after rotation. Rotate only after you have removed active footholds and machine-key harvesters, otherwise the attacker harvests the new key immediately.
My SharePoint server is internal only. Do I still need to act?
Yes, though your urgency is lower. CVE-2026-45659 needs only Site Member permissions, so any compromised employee credential or a foothold elsewhere on the network puts an internal server in reach. Patch it, and put it on a decommissioning plan given the July 14, 2026 end of support for SharePoint Server 2016 and 2019.
How quickly should a private company patch a KEV-listed flaw?
Use the federal deadline as your benchmark. Under BOD 26-04, CISA gave agencies roughly 72 hours - three days - from KEV listing for each flaw in this cluster. Known-exploited vulnerabilities on internet-facing systems should be treated as emergency change, not next month's maintenance window.
Need Help With Cloud Services?
LayerLogix provides expert cloud services solutions for businesses across Houston and nationwide.
Related Articles
Need Expert IT Support?
Let our team help your Houston business with enterprise-grade IT services and cybersecurity solutions.