Demo Request
Take a personalized product tour with a member of our team to see how we can help make your existing security teams and tools more effective within minutes.
Thank you! Your submission has been received!
Oops! Something went wrong while submitting the form.
Home
Blog

Zoomsday: What We Found In Our Customers' Zoom Configurations

Yaniv Blum
Updated
August 30, 2026
August 30, 2026
3 min read
Ready to Close the SaaS Security Gap?
Chat with us

In August 2026, researchers at A Security disclosed what they call the Zoomsday chain: three vulnerabilities in Zoom's screen-share annotation feature, tracked as CVE-2026-53413, CVE-2026-53414, and CVE-2026-53415. Chained together, they let a participant in a Zoom meeting silently take over another attendee's device, or let a presenter take over everyone watching. No click, download, or approval prompt required. 

We checked our own customer base against it and the number of affected customers was higher than we expected.

What the bug did

Zoom's annotation feature lets participants draw, highlight, and add text on a shared screen. To make that work, Zoom runs a proprietary protocol that opens a direct channel between whoever is sharing their screen and everyone viewing it. The chain runs through three separate memory-safety bugs in how that protocol handles annotation data. CVE-2026-53413 is a buffer over-write (CVSS 8.3) in the routine that processes formatting data, in a function called CAnnoFormatBlock::Deserialize. CVE-2026-53414 is a related buffer over-read (CVSS 6.5). CVE-2026-53415 is a use-after-free (CVSS 8.3) that researchers chained with the other two to gain code execution. In a proof of concept on macOS, researchers used the bug to silently launch Safari on a target's machine with no warning and no visible sign anything had happened.

The flaws affected Zoom clients across Windows, macOS, Linux, iOS, and Android. Zoom covered the three in bulletins ZSB-26015, ZSB-26016, and ZSB-26017, and has shipped patches across supported platforms and release branches.

The deeper issue at play

A Security found and weaponized this chain in about 24 hours, using fewer than 20 prompts to a publicly available AI model. The firm's cofounder, Omer Gull, put the shift in blunt terms: work that used to take a team of five people roughly six months now takes a small team an afternoon.

Zoom's annotation protocol has been running in production for years without anyone noticing this. It's a stark example of how fast the floor moves now. The same AI capability that compresses vulnerability research from months to hours cuts both ways: attackers get faster, and so do defenders who build their detection pipeline the same way. We built ours that way. Our posture-check research team uses AI-assisted analysis internally, which is part of why we could ship coverage for this within days of public disclosure instead of weeks.

What we found in our own data

We ran our customer base against the vulnerable Zoom builds. 55% of Zoom users across our U.S. customer base were running a version exposed to Zoomsday. Exposure per tenant ranged from 10% to 100% of users. This wasn't a niche configuration issue. Most organizations run Zoom, and most organizations hadn't gotten to this patch yet.

Reco’s response

Within days of disclosure, our security research team released:

  • Six new posture checks that flag unpatched or vulnerable Zoom configurations.
  • Two alerts for risky setting changes that reopen the attack surface, specifically annotation being re-enabled or screen-share permissions being widened to all participants.
  • Branch-aware version comparison, rather than a flat "is this version old" check. Zoom ships patches across multiple release branches, and a naive version check flags clients that are already safe. Our branch-aware logic cut false positives by 6,044 users compared to that naive approach.

That last point particularly matters because a posture check that floods your team with false alarms gets ignored within a week. One that tells you exactly which of your 32,902 users are actually exposed gets acted on.

The takeaway

Zoomsday is a Zoom story on the surface. Underneath, it's a story we’re telling more and more often about the shrinking gap between a vulnerability being found and it being exploited at scale. It’s about whether your security tooling can keep pace with that gap. AI cut the time to find this chain of bugs from months to hours. It should also be cutting the time it takes you to know you're exposed.

If you're a Reco customer, check your Zoom posture dashboard for the Zoomsday checks now live in your environment. If you're not yet a customer and want to see what your own exposure looks like, reach out and we'll walk you through it.

No items found.

Yaniv Blum

ABOUT THE AUTHOR

Yaniv Blum is a Product Manager at Reco, where he focuses on SaaS identity security and integrations. He brings a strong technical foundation from prior roles in cloud security engineering at Wiz and Dome9 Security, where he worked on security and compliance automation across cloud environments.

Technical Review by:
Gal Nakash
Technical Review by:
Yaniv Blum

Yaniv Blum is a Product Manager at Reco, where he focuses on SaaS identity security and integrations. He brings a strong technical foundation from prior roles in cloud security engineering at Wiz and Dome9 Security, where he worked on security and compliance automation across cloud environments.

Table of Contents
Let’s Talk About Your Non-Human Users
Chat with us
Get the Latest SaaS Security Insights
Subscribe to receive updates on the latest cyber security attacks and trends in SaaS Security.

Explore Related Posts

LWR Is the New Aura: A Pentester's Guide to Salesforce's WebRuntime API
Nitay Bachrach
Salesforce Experience Cloud sites - public-facing web apps that run natively on top of a Salesforce org - are one of the most consistently under-scrutinized parts of the Salesforce attack surface. They're reachable by anyone with a browser, they're driven by a guest user identity that inherits whatever sharing rules and object permissions an admin configured for it, and unlike the rest of the org, nothing about them requires a login to start probing.
Two Frontier Labs, Two Weeks, One Root Cause: What Anthropic's Eval Incident Tells Us About Agent Security
Tal Shapira
On July 30, Anthropic published a retrospective on three cybersecurity evaluation incidents involving its Claude models. It's a striking piece of transparency — and it lands just weeks after OpenAI disclosed something strikingly similar involving Hugging Face.
The “City-Forum” Campaign - An advanced attacker is targeting Salesforce and ServiceNow instances worldwide
Nitay Bachrach
Reco is tracking an ongoing campaign we've named the City-Forum Campaign, after a domain tied to the threat actor's IP (more on that below). A single server is pulling records out of Salesforce Experience Cloud sites and ServiceNow (SNOW) Service Portals, from infrastructure that has been standing since March 2025.
See more featured resources

Your agents are already running. Do you know what they're doing?

Request a demo