Zoom screen‑sharing flaw lets attackers seize devices A single Zoom screen‑share let an attacker hijack any participant’s phone or Mac. Researchers demonstrated remote code execution by exploiting the way Zoom renders…

The proof‑of‑concept required only a handful of commands fed to a public AI tool. The researchers reported that fewer than 20 prompts were enough to locate the flaw and generate a working exploit. The AI‑assisted appr…

AI tools cut the discovery time The AI model used was publicly accessible and designed for code assistance. By iteratively asking the model how Zoom handled screen‑share buffers, the team coaxed it into exposing a mem…

Because the AI answered with concrete code snippets, the researchers avoided the usual trial‑and‑error that can take weeks. The speed of the process raised concerns that similar tools could be weaponized by less‑skill…