Comment by reassess_blind

Yes, it is substantially different. A targeted, relentless attack by a state of the art cybersecurity model is far more likely to find obscure vulnerabilities than a traditional automated attack/fuzzer. These models are so much better at finding security holes than anything we've seen before.