Bottom line: On 2026-11-16 Huntress reported confirmed exploitation of a ScreenConnect client flaw, CVE-2026-84869, against technicians at managed service providers (MSPs). Our assessment: if an MSP has remote access to your environment, the security of that provider’s technician laptops is now part of your own risk, and the fix has to be applied on those laptops, not only on the provider’s server.
What changed since last week
Our 2026-11-16 briefing on remote management tooling noted that Huntress had found more than 2,400 unauthorised guest sessions on over 100 ScreenConnect servers and offered a hypothesis that they were bait for this flaw. The new Huntress post, published 2026-11-16, reports exploitation of the same flaw in other sessions, drawing on server logs and a technician’s workstation. Huntress says these sessions are distinct from its earlier reporting, and its crypto-themed unauthorised sessions were not observed being used this way, so the bait hypothesis remains unconfirmed.
What Huntress reports
CVE-2026-84869 was published on 2026-09-08 with a CVSS score of 9.9. That score comes from the CNA, ConnectWise; NVD shows it as secondary and has no primary score. The CVE record, whose text is ConnectWise’s, says a flaw in the ScreenConnect client may allow files to be transferred and executed through an active remote session without authorisation or host confirmation. It says ScreenConnect servers are not impacted, and NVD lists all versions before 26.6.5 as affected. We did not open the ConnectWise bulletin itself. CISA’s Known Exploited Vulnerabilities (KEV) catalog lists the CVE as added on 2026-09-11.
Per Huntress:
- One full compromise. On 2026-11-12 a technician at one MSP joined a blank-named session. Huntress says a script pushed by the guest ran on the technician’s workstation without a prompt, and about nine seconds later a rebranded ScreenConnect client, giving the attacker remote control, was running and connected to the attacker’s relay.
- A large automated spray. Across 42 MSP ScreenConnect servers between 2026-10-20 and 2026-11-16, Huntress counted 688,000 requests from attacker-controlled guest sessions asking to run batch files, from 29 source IP addresses. The sessions repeat the request in case a technician joins. Huntress says it assesses with high confidence that the spray operator and the operator behind the 2026-11-12 compromise are the same.
- A test run. On 2026-11-13 a guest pushed a harmless text file to a technician at a second MSP, and the technician’s client opened it. Huntress calls this a canary: a check that the exploit works before anything real is sent.
- Targeted attempts. In October, an attacker joined three real customer support sessions as an extra guest and asked the technician’s client to run a file whose path pointed into the Windows Startup folder, which would have dropped a script there if the client honoured the path. Huntress does not say whether any of these succeeded; for the one technician on a Mac the Windows path failed.
- Out of date clients. Huntress says many technicians in these cases were still running 22.x, 24.2 and 25.8 clients weeks after the fix shipped.
Huntress also states that this activity is separate from the SCKit worm, which it says appears to be leveraging a vulnerable third-party extension on the server side and not directly exploiting this flaw.
Our assessment
This is our reading, not Huntress’s.
- The weak point is a person’s laptop, which holds the keys. Huntress notes that a technician workstation typically holds RMM consoles, saved credentials and access to every customer. One compromised laptop can reach many clients.
- Patching the server alone does not help. The CVE record says servers are not impacted. A provider can report “we are up to date” about its server and still have exposed technician clients.
- Proportion. Huntress reports one confirmed full compromise and one confirmed canary in this post. The spray is large, but a request is not a victim. As of 2026-11-17, in the Huntress post, the NVD record and the CISA KEV entry (which lists known ransomware campaign use as “Unknown”), we found no public count of affected customer organisations.
Five questions for your MSP or IT lead
- What is the ScreenConnect client version on every technician device that touches our environment, and is it 26.6.5 or later?
- If patching is not complete, has the
TransferFilespermission, which Huntress relays as ConnectWise’s interim mitigation, been removed from roles and session groups? - Has anyone searched the server history for guest file-run requests that overlapped with a technician being connected, as Huntress describes, and what was found?
- If a technician workstation were compromised, which of our systems and credentials would be reachable, and how fast could the provider rotate them?
- What does our contract require the provider to tell us, and by when, if one of its tools or technician devices is compromised?
What we could not establish
- CISA KEV lists the CVE as exploited from 2026-09-11. The specific MSP observations in this post are Huntress-only as of 2026-11-17 in the sources checked.
- Huntress does not name the attacker’s goal. The decoy session names may be lures, which Huntress describes as a possibility, not an observation.
- We did not independently test the detection logic or the indicators; we cite them as published by Huntress and do not reproduce them here.
References
- Huntress: CVE-2026-84869 in ScreenConnect Exploited to Run Malicious Files on MSP Technician Accounts
- ConnectWise: ScreenConnect security bulletin, 2026-09-08
- NVD: CVE-2026-84869
- CISA Known Exploited Vulnerabilities catalog (JSON feed)
- Huntress: Attackers use Crypto Exchange Names to Disguise Unauthorized Sessions on 100+ ScreenConnect Servers