Zoom Annotation Flaws Could Let Meeting Participants Hijack Attendees’ Clients

Security researchers at A Security, an Israeli-founded offensive-security startup, have disclosed three vulnerabilities in Zoom’s annotation tool that could allow a meeting participant to take over another attendee’s client without any user interaction. The flaws, tracked as CVE-2026-53413, CVE-2026-53414, and CVE-2026-53415, affect Zoom Workplace, Zoom Workplace VDI Client, Zoom Rooms, and Zoom Meeting SDK across multiple platforms.

The vulnerabilities stem from improper handling of annotation data. A drawing is transmitted as a structured object with counts that the receiver trusts to determine how much data to read. One flaw involves a buffer over-write in a fixed 128-byte buffer, while another is a buffer over-read that could leak uninitialized heap memory containing code and vtable pointers, potentially enabling ASLR bypass. The third is a use-after-free issue. Additionally, a missing check on message origin allows a malformed drawing to be sent to all meeting participants, as the dispatcher does not verify the sender’s seat.

Zoom has released patches in June and July 2026, before public disclosure. No exploitation has been reported, and none of the CVEs appear in CISA’s Known Exploited Vulnerabilities catalog. Zoom rates the flaws with CVSS scores of 8.3, 6.5, and 8.3, while A Security assigns all three a score of 9.0 under CVSS 4.0. The vendor also credits its own internal team for one of the flaws.

The researchers claim they developed a working exploit in under a day using fewer than 20 prompts on publicly available AI models, highlighting the growing accessibility of exploit development. This disclosure follows OpenAI’s release of GPT-5.6-Cyber to vetted partners only, underscoring the debate over AI’s role in cybersecurity.