AI Tool Finds Zoom Hijack Bug in Under 20 Prompts

Wired AI · rss · 2026-08-11

Security researchers found that a public AI tool needed fewer than 20 prompts to discover a severe flaw in Zoom's screen-sharing feature. The vulnerability allowed anyone on a call to hijack another participant's device. Zoom has since patched the issue.

Original post →

More from Safety

Safety channel →