Arora advocates virtual patching, a proven technology available from security vendors, as the most obvious interim solution to AI-accelerated vulnerability discovery.

The most obvious solution is virtual patching. This is a time tested technology that is currently available with cyber vendors for endpoints, WAF, cloud, network use cases.

Nikesh Arora says that as model performance improves, vulnerabilities will keep being discovered and cannot all be fixed at once, since mass patching risks unknown infrastructure side effects; he proposes hybrid patching.

If you apply many patches at once, you also don't know what side effects might occur in infrastructure

Arora says that as AI models uncover more vulnerabilities, no single fix can address them all at once.

As AI models find more and more vulnerabilities, there is no single solution that can fix them all at once.

Arora proposes hybrid patching, pairing software fixes with network-layer virtual patches, because mass patching risks unpredictable infrastructure side effects.

If you apply many patches at once, you cannot know what side effects might occur in the infrastructure

Arora argues that defensive agent swarms are impractical due to token costs and current security structures, and that the contest with attackers is about speed, not volume.

Attackers stitch together multiple vulnerabilities, slip in while staying hidden, and extract data quickly. This is not a volume fight, but a speed fight.