I saw Linus Torvalds expressing concern over how…

Per Axbom Akkoma

I saw Linus Torvalds expressing concern over how ”AI-powered” bug hunters have made the Linux security mailing list unmanageable. So I sketched out a quick visualisation of this 😁

https://www.theregister.com/security/2026/05/18/linus-torvalds-says-ai-powered-bug-hunters-have-made-linux-security-mailing-list-almost-entirely-unmanageable/5241633

Bug reports before AI and after AI. Before AI the process was straightforward and simpe, a human found a bug they reported, this was evaluated/replicated and added to the backlog, later being patched/fixed. The human receives a thank you.

After AI, hundreds (if not thousands) people use the same AI software to find the same bug. They all assume they have found the bug and report it, along with all the other hundreds of reports of the same bug, and the ones the core team have already found (maybe even using the same sotware). It’s now a rush to triage this into all the hundreds of reports that have been seen before and already along in the process, things that are not actually bugs and then the ones that may be real bugs. The time to manage the real, new bugs is at an all-time low. It’s not hard to imagine that developers are resorting to vibe coding to fix the bugs.

🔗 Originally posted on Akkoma