Someone is making the tools to find these bugs. It's not like they're telling ChatGPT "go find bugs lol"

And running those models on large codebases like these isnt anywhere close to free either.

Does it matter? Either it's a valid bug or it's not. Either it's of high importance or it's not.

A human at Google investigates all of the bugs fuzzers and AI find manually and manually writes bug reports for upstream with more analysis. They are certainly paid to do that. They are also paid to develop tooling to find bugs.

I'm not sure what you think you mean when you say "running AIs indiscriminately". It's quite expensive to run AI this way, so it needs to be done with very careful consideration.

Still, they are paying for the computing resources needed to run the AI/agents etc.

Someone started it running, they are responsible for the results.

They certainly paid someone to run the so-called AIs.