The lede
WeWorm, a zero-click WeChat worm that AI mostly wrote
Calif Research
Calif's team found a memory-corruption bug in WeChat's VoIP stack and had a working remote-code-execution exploit about two days later, then a self-propagating worm a week after that — a target the size of a billion accounts, compromised by a call the victim never has to answer. The line to sit with is their own account of the division of labor: the humans supplied judgment about what to target and how to test it safely, and AI did the rest. Tencent patched it on August 21 and this went public on the 8th, so the news is not the hole but the schedule.