This situation is temporary.

It is obviously true that there can only be a finite number of serious security bugs in any given software. For a given level of LLM capability, there is a finite subset of these bugs that is actually discoverable and exploitable. Once this finite subset is exhausted, the tide of new vulnerabilities will stop.

Perhaps we will see a few more tides like this, as LLM capabilities advance further. But there will be an end.