Skip to content

Response, New York. Noah Park: Check Point Patches Actively Exploited Management Server Path Traversal. Detection, Dallas. Ava Okello: Report Finds Nearly Half of Deployed Detections Need Attention Before SOCs Can Trust Them.

SOCtember

Always First. Fast SOC News.

Remedio Research title card reading Remote Code Execution in TeamViewer.
Image: Remedio Research (fair use). Title card from Remedio's disclosure, published Oct. 29, 2026.

Tools

Remedio Says a Single 32-Bit Multiply in TeamViewer Let a Malicious Host Run Code on the Support Technician's Machine That Connected to Help It, and Version 15.83 Fixes It

CVE-2026-111585, rated 8.8, sits in the viewer's screen decoder and needs only an accepted session. TeamViewer's Oct. 27 bulletin fixes it alongside two more client flaws, says the company knows of no exploitation and lists patched builds for legacy 13.2, 14.7 and 15.64 installs.

James Whitford, ToolsAustin6 min read

AUSTIN - A malicious or compromised computer on the receiving end of a TeamViewer support session could crash, and in a lab demonstration take over, the technician's machine that connected in to help it, according to research published on Thursday by the security firm Remedio. TeamViewer fixed the flaw, CVE-2026-111585, in version 15.83 and lists it in security bulletin TV-2026-1011, issued on Oct. 27.

The bug inverts the usual threat model for remote support tools, where the worry is an attacker controlling the remote end of a session against a user. Here the attacker is the host that shares its screen, and the victim is the viewer. "You don't have to click anything, and no password is stolen: the whole attack is based on accepting the connection," wrote Omri Dar, who leads vulnerability research at Remedio.

TeamViewer rates the flaw 8.8, high, with a vector that requires user interaction, which Remedio says reflects the victim accepting the session. TeamViewer's bulletin says that at the time of publication the company was "not aware of any active exploitation in the wild."

One multiply, two different sizes

Remedio diagram of the attack: the IT support technician's viewer connects in to help a remote machine, and the attacker-controlled host sends back a malicious screen frame that runs the attacker's code on the technician's computer.
Image: Remedio Research (fair use).The attack vector as drawn in Remedio's disclosure.

When a session starts, the host tells the viewer the width, height and color depth of the screen it is about to stream, and the viewer allocates a decode surface for incoming screen updates. Remedio found that the viewer computes the byte size of that surface in 32-bit arithmetic, stride times (height + 0x10), and only afterward widens the result to 64 bits for the allocation. The product can wrap, so the buffer can be far smaller than the surface the viewer thinks it has. The run-length decoders that write pixels into it check each update against the declared width and height, not against the bytes actually allocated.

Remedio's worked example asks for a 32768 by 32768 surface at 32 bits per pixel. The viewer is handed about 4 MB for a surface that still reports itself as nearly 33,000 rows tall, so every row past the 31st is written out of bounds. "The allocation shrank; the permission to write did not," Remedio wrote. Because the attacker picks the dimensions, it can also pick the buffer size: a width of 0xFFFFFFF5, a height of 1 and 8 bits per pixel wrap the allocation down to a roughly 0x440-byte buffer in a small heap size class, next to objects an attacker can groom.

The host can also steer the viewer onto the vulnerable path. The viewer uses this bitmap surface only when it has no hardware video decoder for the session and otherwise decodes AV1 or H.264, but Remedio said a host that refuses to negotiate the video codecs forces the bitmap path. Nothing on the receive path enforces an upper bound on the dimensions a host announces.

From crash to code execution

Remedio lab screenshot of two Windows machines in a TeamViewer session, the support side on the left and the host on the right, from the opening of its end-to-end demonstration.
Image: Remedio Research (fair use).Opening of Remedio's end-to-end attack demonstration.

On a stock Windows install, Remedio said, the overflow produces a clean, repeatable crash, a remote denial of service over a normal session. To go further, it overwrote a pointer in the corrupted surface object that the viewer later calls through, bypassed Control Flow Guard and DEP, routed the call to WinExec and ran a program of its choosing on a fully updated viewer. Remedio said every link ran on the shipping binaries, with one caveat: in the demonstration, the corrupting bytes are placed by a host-side hook that stands in for the wire tile payload and only for the heap-grooming step.

"An accepted session is all it takes, and that is the product's normal mode of operation," the researchers wrote. Remedio tested the Windows viewer, TeamViewer.exe, at 15.80.6.0 and found it affected, used the Windows host at the same version for delivery, and observed the vulnerable constructor still present in the 15.82.6.0 viewer. It reported the flaw on Aug. 14; TeamViewer accepted it on Oct. 22 and shipped the fix in 15.83 on Oct. 27.

What the bulletin covers

TeamViewer security bulletin TV-2026-1011, Security Update for Vulnerabilities in TeamViewer Clients, issued Oct. 27, 2026, CVSS up to 8.8, with CVE-2026-92372, CVE-2026-92374 and CVE-2026-111585.
Image: TeamViewer (fair use).Bulletin TV-2026-1011, captured Oct. 30, 2026.

TV-2026-1011, last updated on Oct. 28, fixes three client flaws. CVE-2026-111585 affects the Full Client and its Portable version before 15.83 on Windows, Linux and macOS, legacy 15.64, 14.7 and 13.2 builds, and the Android Remote Control app before 15.83. TeamViewer describes the impact as denial of service or, "depending on the platform," potential code execution with the privileges of the viewer process.

The second, CVE-2026-92372, also rated 8.8, is an out-of-bounds write in the screen-image decoder of the Full Client for Linux and macOS before 15.83. TeamViewer says it can be triggered with crafted screen-image data during a remote-control session or by persuading a user to open a crafted TeamViewer session recording, a .tvs file. The third, CVE-2026-92374, rated 7.8, is a time-of-check time-of-use race in the macOS Full Client and Host privileged helper that lets a low-privileged local user install a malicious package and gain root.

TeamViewer's only listed mitigation is to update to 15.83 or later. For machines that cannot move to the current release, the bulletin lists fixed legacy builds: 15.64.10 for Windows 7 and 8; 14.7.48868 on Windows, Linux and macOS; and 13.2.36231 on Windows, 13.2.154008 on Linux and 13.2.154007 on macOS.

The tooling takeaway

For a SOC or IT operations team, the exposed population is the viewer side: help desk, MSP and administrator workstations that open sessions into machines they do not control. Those are often among the most privileged endpoints in an organization. Our reading of the bulletin is that inventory should cover the Portable build and the Android app as well as installed clients, since both are in scope for CVE-2026-111585 and a portable copy will not show up in every software inventory. Until a viewer is on a fixed build, sessions into unmanaged or untrusted hosts and .tvs recordings from outside sources carry the risk described here.

Remedio frames the lesson for developers in general terms: wherever a size is computed once for an allocation and bounds are enforced against a separately stored dimension, confirm the two can never disagree. It said it found the same class of bug in Adobe Acrobat, tracked as CVE-2026-81983.

Sources:


James Whitford covers SOC tooling, SIEM, SOAR, and detection platforms for SOCtember from Austin.

Related stories

Tools desk