‘Zoomsday’ hack uncovered using fewer than 20 AI prompts

Zoom has patched a large information vulnerability that could let an attacker to hijack anyone's instrumentality during a meeting. In a blog station connected Tuesday, researchers astatine A Security accidental they uncovered the flaw utilizing "fewer than 20 prompts connected publically disposable AI models," as reported earlier by Wired.

The exploit progressive Zoom's annotation feature, which allows users to gully connected their surface portion sharing it with different gathering participants. With the exploit, an attacker could articulation oregon big a gathering and tally malicious codification connected victims' devices, allowing them to bargain data, crook connected the camera oregon microphone, oregon instal malware. The onslaught required nary enactment …

Read the afloat communicative astatine The Verge.