Zoom has patched four security vulnerabilities across its applications, including two that allowed attackers to execute malicious code on the computers of all meeting participants without requiring any clicks, downloads, or other actions from victims. The flaws affect Zoom client applications, VDI clients, Zoom Rooms, and the Meeting SDK, and were disclosed in a report by security firm A Security. Most alarmingly, a single researcher at the firm discovered and built a working exploit for the vulnerabilities using publicly available AI models in under 24 hours with fewer than 20 prompts.
Three of the vulnerabilities are memory corruption issues in Zoom's text annotation feature and affect all Zoom client applications for all platforms before versions 7.1.5 and 7.0.6. Products including Zoom Rooms and Zoom Meeting SDK before version 7.1.0 are also impacted. A fourth flaw affects Zoom Workplace VDI Client for Windows and VDI Plugins on all supported platforms before versions 7.0.11 and 6.6.15. The annotation flaws include a buffer overflow vulnerability (CVE-2026-53413) and a use-after-free memory error (CVE-2026-53415), both capable of enabling remote code execution, plus a missing bounds check (CVE-2026-53414) that can trigger a denial-of-service condition. The VDI flaw (CVE-2026-53416) is a path traversal issue that could expose sensitive information. Zoom is used by 70% of Fortune 100 companies, most Fortune 500 firms, and federal agencies.
According to A Security, "This class of capability would previously have only been available to nation-state threat actors, but the model requiring elite teams, months of effort, and weapons-grade budgets has collapsed." The firm emphasized that "a single researcher was able to develop a nation-state-level exploit in less than a day." The researchers highlighted the enormous blast radius of such an exploit given Zoom's widespread adoption across major corporations and government entities. The vulnerabilities execute silently with no indication that an attacker has compromised a participant's system.
The flaws stem from how Zoom handles annotation data when participants draw, write, or highlight content on shared screens or whiteboards. When a user performs one of these actions, their Zoom client creates a typed in-memory object describing the action, serializes it into a byte stream, and sends it to Zoom's Multimedia Router, which forwards it to all meeting participants whose clients then deserialize the object. Zoom allocates four fixed 128-byte buffers for the deserialized annotation packets but only verifies that packets are non-zero, not their actual size. An attacker can generate a packet that fills and overflows these buffers, inserting malicious code into the application's memory. Data serialization and deserialization operations are a common source of memory corruption vulnerabilities because they're easy to implement incorrectly and the input comes from attacker-controlled sources.
Organizations can protect themselves by updating Zoom clients immediately or by disabling end-to-end encryption for meetings, which allows Zoom's server-side mitigation to filter malicious annotation messages. Administrators can also set minimum version requirements in meeting preferences to block unpatched clients from joining. The researchers recommend using access controls like waiting rooms, passcodes, and authenticated-users-only settings, and disabling unused features including annotation, file transfer, whiteboarding, remote control, third-party apps, and limiting screen sharing to hosts only. The speed and ease with which AI enabled a single researcher to develop what was once nation-state-level exploit capability signals a fundamental shift in the threat landscape. Organizations must assume that vulnerabilities discoverable by AI will be found and weaponized faster than traditional patch cycles can respond, demanding more aggressive feature lockdowns and zero-trust architectures regardless of vendor reputation.