7 Comments

Thanks for putting your vision on the future of security into words Tyler! Very interesting read.

What you are describing sounds as the security holy grail to me: see across the whole environment, find the issues, prioritise, get to the root cause and finally fix them automatically.

That sounds really great!

Then I use copilot/chatgpt in my daily work, and find myself arguing with it about some obscure (in the best case) or obvious thing it missed, which changes the whole dynamic of an issue. 
Maybe the tool is good and I’m just using it wrong. Maybe we’re on a power law with AI capabilities and I’m not able to extrapolate well. I’m not sure.

To be clear, I’m not arguing AI/LLM’s aren’t going to be the thing in cybersecurity for the next few years. But what you’re describing does seem like a lot to ask of it.

Especially with how it seems to have the tendency to be right in 90% of the cases, but spectacularly wrong in 10% of the cases.

Instead, I think it’s an additional tool in our arsenal, to be used on top of our other tooling. We have potential security issues where we don’t have any context (for example: a code linter looking at string concatenation in sql queries), where the fix is so cheap/quick/easy, it is a no brainer.

I wouldn’t start relying on a finicky AI, if I have a stupid tool which works consistently in 100% of the cases.

So anyway, this comment turned out to be a “right tool for the job” manifest… :-) I’m excited to see what the future holds!

Expand full comment

What a great and thoughtful reply! Thank you so much. I agree with you completely .. for today. The idea I'm primarily trying to get across is that if AI continues to advance at an exponential rate in capabilities then this is the natural outcome for the cybersecurity industry as a whole. We will see the next 5-10 years as consolidation of tools and in reality a consolidation of the data set that the overarching AI systems have access to. PANW has already begun to make this play and it will continue to be a battle MSFT, PANW, and one or two others over the next decade. I really appreciate your comments, you are spot on.. for the moment :)

Expand full comment

Couldn’t agree more. A new standard will need to emerge. Elimination of AuditBias requires a standard like the Zerobias - cyber, digital & risk assessment standard. https://www.linkedin.com/groups/14345270

Expand full comment

I enjoyed this post. I feel optimistic that Microsoft's Security Copilot may end up being a good example of some of the progression you've highlighted here.

Expand full comment

Thank you Patrick! I appreciate the kind words.

Expand full comment

I'm glad I've discovered, and now subscribed to, your publication.

Expand full comment

Ant Group, a Chinese financial technology company, is developing a framework called Aspect-Oriented Security, and it plays a role in many real network environments to provide enterprises with unique and accurate contextual information for security observations. It seems that everyone is moving towards the same cybersecurity architecture goal. Hope to discuss it with you all. 814211866@qq.com

Expand full comment