‘Zoomsday’ hack uncovered using fewer than 20 AI prompts

Researchers at A Security used fewer than 20 AI prompts on public models to find a Zoom vulnerability in its annotation feature that allowed device hijacking without user action. Zoom has since patched this flaw affecting multiple platforms.