HomeServicesManaged IT ServicesAI DeploymentAccredOneInsightsAnswersAboutFree Consultation
Remote Support Security

ScreenConnect Attacks: What MSPs Should Do, and a Safer Model

Status snapshot: November 12, 2026. This is a developing incident and details are still changing. Facts below come from ConnectWise and security researchers linked in the Sources section. We update this page as ConnectWise publishes more.

On November 11, 2026, ConnectWise said it was investigating unauthorized access to ScreenConnect partner accounts and the deployment of unauthorized remote-access software in customer environments. Security researchers describe a fast, automated campaign. If you run remote support for other people's computers, this one is worth an hour of your day.

This article covers what is known, what to check today, the questions every MSP and IT team should ask of any remote support tool, and how Accred Connect approaches the same job differently. We build Connect, so read that part with that in mind. We have tried to keep the facts and the pitch clearly separate.

What happened

  • Admin consoles were hijacked. ConnectWise reported that attackers compromised the admin consoles of at least six managed service providers that use cloud-hosted ScreenConnect.
  • Malware went out to managed devices. Researchers at Huntress and others reported a custom toolkit called SCKit pushed to at least 288 devices across 48 organizations, using the console's remote command feature to run scripts on many endpoints at once.
  • The scripts installed a second remote-access client. Reports say the scripts installed an attacker-controlled ScreenConnect client, added Microsoft Defender exclusions and created a scheduled task to replace the legitimate client.
  • Third-party extensions are the leading suspect. ConnectWise disabled five third-party extensions across about 400 partner environments as a precaution. Researchers pointed to a possible authentication weakness in one extension family. ConnectWise says it has not yet confirmed the initial access method.
  • ConnectWise says its own systems were not breached. Its advisory reports no evidence of a compromise of ConnectWise infrastructure and no vulnerability in ScreenConnect itself.
  • Customers on their own servers have to act themselves. ConnectWise can disable the extensions in its cloud, but partners running ScreenConnect on-premises have to turn them off on their own servers.

We are not going to speculate beyond that. The practical point is clear enough: once an attacker holds a remote-support console, they hold a way to run code, as an administrator, on every computer it manages.

Why remote support tools are such a good target

A remote support platform is built to do exactly what an attacker wants: reach a computer without anyone sitting at it, run commands with high privileges, and do it across hundreds of machines from one place. That is useful for IT and valuable for criminals. Three properties decide how bad a compromise gets:

  1. Blast radius. If one login or one console can run commands on every device in every client, one stolen login becomes every client's problem.
  2. Silent execution. If commands run without the person at the computer or a second technician ever seeing them, there is nothing to stop a bad one.
  3. Weak records. If the log can be edited, or does not capture each command and who approved it, you find out what happened late and incompletely.

What to do today if you use ScreenConnect

This checklist follows the guidance ConnectWise, Huntress and Sophos have published so far. Read their advisories for the current indicators.

  1. Find and disable the affected extensions, including on on-premises servers, and keep them off until ConnectWise says otherwise.
  2. Review command history for remote commands or client installs nobody on your team ran. Look for downloads of ScreenConnect installers by PowerShell or curl followed by an MSI install.
  3. Check the server thumbprint on your agents. Every ScreenConnect agent is tied to a server by a public key thumbprint. An agent pointing at a thumbprint that is not yours is not yours.
  4. Inspect endpoints for new Microsoft Defender exclusions and unfamiliar scheduled tasks.
  5. Revoke sessions, reset technician passwords and enforce MFA on every technician account.
  6. Isolate suspect devices and avoid interactive ScreenConnect sessions into them. Investigate with a separate security tool.
  7. Tell your clients. If any of their computers were touched, they will hear it from you sooner or from someone else later.

Questions to ask of any remote support tool

Whatever you run, these questions separate a design that limits damage from one that does not.

  • Can one compromised technician login run commands on every computer without anyone else approving?
  • Is the person at the computer asked first, and can they see and stop what is happening?
  • Are commands classified by risk, and do the risky ones wait for a human?
  • Does every command, approval and result land in a log that cannot be quietly edited?
  • Can the organization switch the tool's powers off in one place?
  • What does the vendor say about encryption, plainly, and does the marketing match the engineering?

How Accred Connect handles remote support

Accred Connect is the remote support and AI-assisted IT support app in AccredOne. It works on Windows PCs and Macs. Here is how it answers the questions above.

Patent-pending agentic IT support

Connect has patent-pending agentic IT support. Hand your AI assistant a ticket and, through Connect, it finds the person's computer, runs diagnostics, picks a reviewed fix, checks that the fix worked and writes it up. A technician stays in charge of anything risky. Because the work is structured, every step is captured, which is the opposite of an unlogged shell session.

Nothing runs without a yes

Nothing runs unless your organization allows it and your technician's rules allow it. The person at the computer is asked first, unless the computer is approved for unattended help, and every step is in the audit log. Consent prompts default to Deny, and no answer within 60 seconds counts as no.

Every command is risk-classified

The assistant's commands are classified before they run. Read-only checks can proceed. Changes follow the policy your organization sets. High-risk jobs wait for a technician to approve them in the console. The assistant cannot approve its own fixes. A built-in deny list blocks access to secret stores such as credential stores and saved Wi-Fi keys, and command output passes through a redactor before the assistant sees it. Where possible, a fix can be undone in one step.

The person can see and stop it

While an AI assistant works on a computer, a session bar at the top of the screen shows that it is active and what it is doing, with a Stop button. Local opt-out and a kill switch for the whole organization sit in the assist policy.

An audit trail you can read

Every request, policy decision, approval, result and consent answer is written to an append-only audit log. When something goes wrong, you can see which command ran, who approved it and what it returned.

Access follows your Hub role

Connect sits behind your AccredOne sign-in. Owners and Admins manage devices and installers, other members see inventory, and the assistant can do what your account can do and nothing more. Enrollment installers are tied to one client and location and can be revoked. Remote-control sessions are designed around short-lived signed credentials tied to one organization, device, session and role, and production traffic requires TLS.

On the encryption question, honestly

People asking about remote support right now often ask whether it is end-to-end encrypted. Connect traffic uses TLS. It is not end-to-end encrypted today, and we do not claim it is. We would rather tell you what exists than have you find the gap after an incident. When end-to-end encryption ships, we will say so here.

Help where people already ask

People can mention Accred IT help in a Teams channel or a linked Slack channel and describe the problem in their own words. It finds their computer, runs safe checks, answers in plain words and hands your technician the notes, a Consult ticket and an alert. Fixes you choose still wait for a technician's OK.

Technicians can work together, and not interrupt people

Several technicians can join one computer, see who is connected, pass control back and forth, and let Owners and Admins end a session. On Windows, Accred Shadow, part of Connect, gives a technician a private workspace on the same computer so they can fix the machine without taking over the person's screen.

Works with the AI assistant you already use

Connect is part of Accred's MCP server, one server for the whole AccredOne suite on the open Model Context Protocol. It works with Claude today, and with other AI assistants that support remote MCP connectors with OAuth sign-in.

What Connect is not

We would rather be plain. Connect does not make an attack impossible, and we are not claiming it is more secure than any specific competitor's product. We have not tested ScreenConnect and make no claims about its internals. What we can say is how Connect is designed: limit what one login can do, ask the person, classify the risk, make a human approve what is dangerous, and write it all down. The Windows installers are code-signed. The Mac installer is not yet notarized by Apple, so macOS blocks it the first time and the person allows it in System Settings, under Privacy & Security.

Price and how to try it

Accred Connect comes with Consult + Connect seats at $29.99 per seat per month, with a 30-day free trial and no card. Create your organization in Accred Hub, and an Owner or Admin turns Connect on. See Accred Connect or read how patent-pending agentic IT support works step by step.

Common questions

Was ScreenConnect itself hacked?

ConnectWise says it has found no evidence that its own infrastructure was compromised and no vulnerability in ScreenConnect itself. Reports say attackers took over the admin consoles of at least six MSPs using cloud-hosted ScreenConnect, and that third-party extensions are the leading suspect. ConnectWise says its investigation is ongoing.

What should I do if my MSP uses ScreenConnect?

Ask them in writing whether they ran any of the affected extensions, whether they have checked their command history and client deployments, and whether technician logins have been reset and protected with MFA. Then ask which of your devices, if any, received unexpected commands.

Is Accred Connect immune to this kind of attack?

No tool is immune, and we do not claim that. Connect is built so that one compromised login cannot silently run risky commands across every computer: high-risk fixes wait for a technician approval, the assistant cannot approve its own work, the person at the computer is asked first unless the machine is approved for unattended help, and every step is in an audit log.

Is Accred Connect end-to-end encrypted?

Not today. Connect traffic uses TLS, and we do not market it as end-to-end encrypted. End-to-end encryption is on our roadmap, and we will say so plainly when it ships.

What does Accred Connect cost?

Connect comes with Consult + Connect seats at $29.99 per seat per month, with a 30-day free trial and no card. It supports Windows PCs and Macs.

Sources

Want help checking your environment or moving remote support? Accred's managed IT team can review your remote access tools, set up Connect, the AI assist policy and who approves what. Talk to us.