Question AI future

[PitDuran]

Hi everyone,


I’ve been following this thread about Hard Configurator and Windows hardening tools, and it got me thinking about how modern technologies like AI automation could enhance or change the way we secure systems. What are the potential benefits and risks of adding AI automation to system hardening?

 

[bazang]

AI will eventually kill us (the hoomans) all.

 

[Morro]

AI will eventually kill us (the hoomans) all.

Oh great, thank you very much, now I constantly hear the main music from the Terminator in my head.

 

[bazang]

Oh great, thank you very much, now I constantly hear the main music from the Terminator in my head.

It won't be AI killing us all via "Rise of the Machines" or Elysium or Total Recall style robots.

 

[Morro]

It won't be AI killing us all via "Rise of the Machines" or Elysium or Total Recall style robots.

Oh ok, then I am good ... for now.

 

[Victor M]

AI automation

AI is AI, I understand that. But what sort of automation are you thinking of ?

 

[Divergent]

Hi everyone,


I’ve been following this thread about Hard Configurator and Windows hardening tools, and it got me thinking about how modern technologies like AI automation could enhance or change the way we secure systems. What are the potential benefits and risks of adding AI automation to system hardening?

In 2026, AI automation is no longer a strategic advantage, it’s a survival requirement against agentic threats. While AI-driven remediation is essential for speed, its non-deterministic nature introduces risks like prompt injection and 'Shadow Automation.' To secure the modern enterprise, we must anchor our Continuous Exposure Management (CEM) with rigorous human governance to ensure our defenses don't become our newest vulnerabilities.

 

[stonjean633]

“AI won’t replace humans. But humans who use AI will replace those who don’t.” Sam Altman

 

[bazang]

AI/LLM/ML - even with solid instruction sets - can still be:

1. Incorrect
2. Incomplete
3. Needlessly restate things over-and-over
4. Add a bunch of extraneous or superfluous information
5. Waste hours and hours of a peoples' valuable time

I am certain there is a long list of "more."

Somebody goes ask multiple AI platforms and consolidate everything that each AI platform returned/stated.

@Divergent

 

[Halp2001]

If AI ever kills us, it probably won’t be with red lasers like in Terminator, but with a massive blue screen crash and an update that never ends. The paradox is that the same “savior automation” could turn into the bug that knocks us out of the game… and at that point, not even Skynet would bother to laugh.

That said, adding AI automation to system hardening is a double‑edged sword: it can speed up defenses and reduce reaction times, but if misused it may also automate new vulnerabilities, among other things.

 

[Divergent]

AI/LLM/ML - even with solid instruction sets - can still be:

1. Incorrect
2. Incomplete
3. Needlessly restate things over-and-over
4. Add a bunch of extraneous or superfluous information
5. Waste hours and hours of a peoples' valuable time

I am certain there is a long list of "more."

Somebody goes ask multiple AI platforms and consolidate everything that each AI platform returned/stated.

@Divergent

Think of AI like a world-class researcher with no keys to private libraries. It’s incredibly accurate within the public record, but its performance is strictly capped by the quality of its sources and the clarity of your 'research brief.'

 

[Victor M]

@PitDuran Didn't read your original post fully. Particular to hardening; how much faster do you want than checkmarking a few items and hitting OK button? (as in the case of H_C or WHH Light ) . If you are thinking of automatic responses to an intrusion, and I am thinking of the traditional response like forensics, then re-image the drive, then add prevention to stop it from coming around again. I don't know if those 3 steps can be automated. Maybe I just lack imagination because I'm not in my 20's anymore.

EDIT : or maybe it can be automated. Machine Learning can run some recognition task from the decompiled attack binary, and then re-image the drive, then pin point specific parameters that doesn't change in the attack and exfiltrate steps and close some firewall ports. Maybe even disable some OS component being attacked if we don't really need it. That's if we can recover the binary - good hackers won't leave them lying around.

 

[wayner.d]

The thread makes some solid points about both the promise and the risks of AI in system hardening. AI can speed up repetitive security tasks, but it also introduces new challenges if used without careful governance. It’s worth exploring AI automation responsibly and even seeking guidance like ai consulting services (e.g., AI Consulting Services. Artificial Intelligence Consulting Company | ARTJOKER Firm) to balance innovation and safety.