Am I reading the thread correctly? This fuzzer found a "new" issue that was already identified in 2024? This matches what I've heard before about GenAI reporting known errors likely because they were trained on the internet as it's data source.
With a history like ffmpegs, it's just very likely that fuzzer findings have already been reported once. That the fuzzer is vibecoded is kinda irrelevant
7 Comments
Ch3rry314@piefed.social · 41 pts · 5d
Am I reading the thread correctly? This fuzzer found a "new" issue that was already identified in 2024? This matches what I've heard before about GenAI reporting known errors likely because they were trained on the internet as it's data source.
Brownboy13@lemmy.world · 9 pts · 5d
This fuzzer doesn't seem to be an AI/LLM based tool.
technocrit@lemmy.dbzer0.com · 7 pts · 5d
No bug discovered. No "vibecoding" detected.
Gotta love tech "journalism".
sxh3mr@discuss.tchncs.de · 6 pts · 5d
With a history like ffmpegs, it's just very likely that fuzzer findings have already been reported once. That the fuzzer is vibecoded is kinda irrelevant
plateee@piefed.social · 5 pts · 5d
In theory they can "find" new things, but typically it's based on other training data.
Reality@feddit.org · 1 pts · 5d
Do you mean LLMs can find new bug classes or new bugs?
plateee@piefed.social · 4 pts · 5d
New bugs, not bug classes.
They also seem to do well with chaining low severity vulnerabilities to result in an overall high severity exploit.
morto@piefed.social · 1 pts · 5d