On this page (7 sections)
CVE-2026-102764 is a critical flaw in BeyondTrust Remote Support, an appliance used for remote customer support sessions. The GitHub advisory says internal components did not adequately validate the origin of the messages they received, which under the right conditions could let an attacker perform file operations on the appliance.
On October 30, 2026, Rapid7 published an analysis showing that the flaw can be chained into unauthenticated remote code execution, and says a Metasploit module now implements the chain. If you run a BeyondTrust Remote Support appliance that is reachable from the internet, treat this as urgent.
What happened
The flaw was published to the National Vulnerability Database on October 15, 2026, and the record references BeyondTrust's vendor advisory BT26-05. The vendor's knowledge base article is tagged as requiring permissions, so this article does not repeat its contents. Rapid7's later write-up is the main public technical source and goes well beyond the short CVE description.
According to Rapid7, an attacker needs no account and no user interaction. The entry point is the public customer connection used to start a support session.
Why the flaw matters: a plain-language explanation
The weakness class is origin validation. Inside the appliance, components pass messages to each other through an internal router, and some components decide whether to trust a request based on where it claims to come from. Rapid7 reports that a message arriving over the network could claim a reserved local source address, the router forwarded it, and an appliance-side customer worker accepted that claim as proof the request came from a local component.
That trust gives the sender access to an internal file-transfer service. Rapid7 states the attacker controls both the content of an uploaded file and its destination path, subject to the worker's filesystem permissions. In its test, the same forgery also allowed retrieving a file from the appliance.
From file write to code execution
Rapid7 reports that on the tested appliance the file service could write into the web server's public directory. Placing a script file there and requesting it gave code execution as the appliance's site user (UID 1000 in the Metasploit module). In other words, a file-write bug became unauthenticated remote code execution. We are deliberately not reproducing the message-level details here.
Who is affected
The full affected range is unknown. Rapid7 confirmed the issue on Remote Support 24.1.2 (build 1524046) with public issue-menu session starts and thin chat enabled, which it describes as the default. It did not establish other affected releases or fixed builds, and the GitHub advisory lists both affected and patched versions as unknown.
Do not assume an older or newer version is safe. Rapid7 also notes that the appliance's base Secure Remote Access version (7.1.0 on its test VM) is separate from the Remote Support application version, so check the right number when comparing against vendor guidance.
- Organizations hosting Remote Support on-premises or in their own cloud account, particularly where the customer-facing session endpoint is public.
- Website owners whose hosting or agency partner uses BeyondTrust. You cannot patch it yourself, but you can ask whether they have.
How to check your exposure
Start by identifying every Remote Support appliance you own and whether it answers on the internet. Then record the exact application version and compare it with the fixed builds listed in the vendor advisory.
- Find the appliance's public hostname and confirm which of your DNS records point to it.
- Log in to the appliance administration interface and note the Remote Support application version and build, not the base appliance version.
- Open the BeyondTrust advisory BT26-05 and the knowledge base article (permissions required) and compare your build to the fixed releases.
- Check whether public session starts are enabled and reachable from outside your network, since that is the entry point Rapid7 describes.
What to do now
- Read BeyondTrust advisory BT26-05 and apply the update it specifies for your Remote Support version.
- If you cannot patch immediately, restrict network access to the appliance, for example with firewall rules that limit the customer session endpoint to the traffic you actually need.
- Review whether public session starts are required. If they are not, ask BeyondTrust support how to limit exposure.
- Check the appliance for unexpected files, especially recently created script files in its web-accessible directories, and review web and application logs for unusual requests to the session start and WebSocket endpoints.
- If you find signs of tampering, treat the appliance as compromised: isolate it, preserve logs, and rotate credentials and secrets that the appliance could access.
- Confirm that downstream systems reachable from the appliance use separate, least-privilege credentials.
- Ask any hosting or agency partner that uses BeyondTrust whether they have applied the fix.
Detection and monitoring
Because a Metasploit module exists, opportunistic scanning is plausible, although no source in our research confirms exploitation in the wild. The CVE record references a CISA Known Exploited Vulnerabilities entry, but the data we have shows no listing, so check the CISA catalog entry for the current status.
- Alert on new or modified files in the appliance's web root and on processes spawned by the web or application user.
- Watch for unexpected outbound connections from the appliance.
- Keep an eye on the vendor advisory and the NVD record for updated affected and fixed version data, since both are incomplete today.
Webforta's certificate and DNS monitoring can help you notice unexpected changes on the public hostname of an appliance, and the free SSL checker is a quick way to see what certificate the endpoint presents. Neither replaces patching or checking the appliance itself.
What we still do not know
- The complete list of affected versions and the exact fixed builds. The GitHub advisory lists them as unknown and Rapid7 tested only one build.
- Whether the flaw is being exploited in real attacks. The research contains no confirmation either way.
- Which configurations beyond the default tested by Rapid7 are exposed.
For broader guidance on reducing risk across the systems that support your site, see our guide to small business website security.
Sources
This article is based on the public sources below. Details can change as vendors and researchers publish more; check the original advisories before acting on version numbers.
