Zoom Screen-Sharing Flaw Cracked With Fewer Than 20 AI Prompts

A camera was nearly hijacked mid-video-call

A vulnerability was discovered in Zoom that could let an attacker seize control of another participant's laptop camera and microphone simply by joining a meeting. Without any special hacking skills, an attacker who merely hosted or joined a call could run malicious code on a victim's device, steal data, secretly activate the camera and microphone, or plant additional malware. With video conferencing tools now a default part of work, simply opening a meeting window could have become a risk in itself.

A drawing tool became the entry point

The vulnerable pathway was Zoom's "annotation" feature — the common collaboration tool that lets users draw underlines or circles on a shared screen. Security research team A Security found a flaw in how this feature was processed, discovering that an attacker could gain command execution privileges on another participant's device without any additional manipulation. Because screen sharing is a feature already built on trust, the flaw effectively broke that very trust relationship.

A flaw found with just a few words to an AI model

What stands out about this discovery is the method used. In a blog post last Tuesday, A Security said it found the flaw "using fewer than 20 prompts to a publicly available AI model." Compared to the traditional approach of combing through code for weeks to find a single vulnerability, this means the flaw's location was pinpointed through just a handful of conversational queries. The story was first reported by Wired and later picked up by The Verge.

AI is being used on both sides

This case fits into a pattern that has emerged over the past few weeks. On August 10, OpenAI introduced GPT-5.6-Cyber, a model dedicated to cybersecurity, saying it had used the model to uncover previously unknown vulnerabilities in widely used open-source software such as Chrome's V8 engine — a case of AI being formally deployed for defensive vulnerability discovery. Around the same time, an Australian developer's Claude-based agent drew attention after it independently found an authentication flaw in a gym booking system and used it to manipulate booking rankings. Both cases show that AI's ability to read code and identify logical flaws has reached a level usable even by non-developers.

This trend means the same tools are available to both defenders and attackers alike. If a security research team can find a vulnerability using fewer than 20 AI prompts, the concern follows that those with malicious intent could find flaws at a similar pace. Some point to the recent LiteLLM supply-chain attack, which exposed credentials from 2,500 organizations, as evidence that the pace at which security incidents occur is itself accelerating in the age of AI.

OpenAI unveils GPT-5.6-Cyber, a model dedicated to cybersecurity (/2026/8/we-re-expanding-our-cybersecurity-initiative-daybreak-and-introducing-gp)

So what changes now

There is no immediate action Zoom users need to take. The vulnerability has already been patched, and no reports of actual exploitation have been confirmed. Still, the incident can be read as a signal that the way software security is verified is changing. Flaws that once required manual work by professional penetration testers can now be discovered by anyone asking a few questions of a publicly accessible AI model. It adds one more reason for companies to make AI-based vulnerability scanning a standard step before releasing their own software.