Post #1998044
2026-04-30 18:01 UTC
@diazona
There's a lot of AI slop throughout everything. It's a real exploit, of course, but everything surrounding it and their incredibly hamfisted disclosure really give me the vibe of some guys who threw an LLM with a fuzzer at the kernel and had the LLM write it all up.
I assume we'll be seeing a lot more of this type of disclosure slop. :(
Replies (1)
-
@aoanla@hachyderm.io 2026-05-01 02:15
@jbowen @diazona LLM with a fuzzer and some known "soft" parts in the kernel - the (poor) write up admits that the prompt that found this vuln told the tool the key function to try to exploit (although the exploits were all generated by the tool - which also explains why the original code *works* but also has bugs like claiming to support multiple payloads via arguments passed to it, but in fact does not).