AI-found bugs aren't proving any easier to exploit despite the hype (theregister.com)

I was under the impression the 'dynamic automation' of chaining these sorts of exploits was the real concern?