Securing the Future: Microsoft's Integration of ChatGPT in Cybersecurity

0:00If you follow the podcast for a while you'll know that I've run a bunch of series that essentially are you know how ChatGPTGPT will disrupt travel or how ChatGPT will disrupt X, Y and Z industry I've done it for a bunch of things. And it's kind of interesting because recently I feel like in the last week or two we've really seen especially with chat GPT plugins The knob turned up on directly. It's not I'm not like Hypothesizing anymore about how chat GPT could theoretically disrupt an industry A lot of the tools are now coming out and one of the big tools that I want to talk about today is Cyber security.

0:31So, you know, we've talked about how In the past about like how chat GPT, you know a lot of it got a lot of criticism essentially by a bunch of people being like oh you know people can use it for elaborate phishing attacks or hacking or other things and of course open ai had to come up with a bunch of security measures so that people weren't like doing that with it essentially but what is really interesting is it is now being used actively to thwart uh cyber security hacking um and a lot of those kind of threats so today on the podcast we're going to jump into what is happening right now and the biggest thing is that microsoft is bringing open ai's chat gpt capabilities to its own cybersecurity business.

1:07So Microsoft has about a$20 billion cybersecurity business called Microsoft Security Copilot. Well, I guess that's their new one that they've just launched is called Copilot, but Microsoft Security has been out for a while and now they're integrating this into it. And you know, you've seen the same thing. Microsoft obviously got the super in with OpenAI when it gave them$10 billion. And so we saw Microsoft integrate ChatGPT into, you know, Word and PowerPoint and Excel and a lot of different new tools they come out with and Bing and stuff. And so now the latest thing they're doing is they are, is that Microsoft is bringing ChatGPT's technology into cybersecurity specifically to help people defend in a way that they've never been able to defend before, as Vasu Jekal, who is Microsoft's CVP of security and compliance, said.

1:58So he essentially said that you're now going to be able to protect and disrupted attacks when they're happening and this is all going to be used um is all going to be done through gpt4 um and so what's and okay so also the other thing to mention though it's going to be gpt4 and then also microsoft's own kind of security specific model that they've trained um so this is similar to what we've seen with github and microsoft's uh copilot for coding right so they have this like you know github github copilot that helps you write code and it was trained off of GitHub's data set in order to, you know, make it a really good model that can write code.

2:37So they're doing something similar because Microsoft has been doing the cybersecurity stuff for a while, right? They have a$20 billion business here. They essentially have been able to help train a specific model based off of a lot of that info, which is, you know, interesting how they acquired that. I'm sure there's terms of service and anonymity in the data they've collected from companies or people that they work for. But that's another, that's another thing. But they've been able to train their own model specifically on doing that and this is exactly I guess this is kind of what it does so essentially it's going to be an AI bot that allows cybersecurity professionals to do things like quickly pull together information on the latest security incidents in their company and also to kind of dig into potential threats and even quickly look at data on common vulnerabilities and exposures that they might have so one one example that Microsoft showed in their presentation on this whole thing.

3:29They showed how the security copilot can look into a cyber attack to pick apart how the hacker got into a network and onto a victim's device in the first place, right? So if they know what the vulnerability is, obviously they know what they can patch and do to prevent it in the future. They said it's the first and only generative AI based model that is out there and it's kind of a one of a kind. So it's never actually happened before. So I think Microsoft is actually getting a lot of benefits by being the first to add these AI tools, right? There's people that are trying to catch up, right? Google Bart is trying to catch up to Bing, and Google's trying to integrate with their own Word and Sheets and different document things in order to try to catch up to Microsoft.

4:14But Microsoft is launching first, and so a lot of these people are playing catch up. So as I said before, Microsoft already sells a bunch of different cybersecurity offerings and their kind of cybersecurity business. They have Microsoft Defender, Microsoft Entree, Microsoft Purview, and Microsoft Sentinel. And so that's kind of their whole cybersecurity arm, which is$20 billion a year. But what's interesting is all of those different product offerings were used to help train kind of their specialty models. So essentially, it's going to continually improve as it learns from a company's own data.

4:52but the thing that they've said and they've actually said this for other people as well and I think people are going to find this really important the data that is being given to Microsoft Security Copilot is never going to be used to teach the broader Copilot algorithm so essentially what that means is that a customer's information is going to remain its own so if you know Microsoft Security Copilot is sitting there taking your data and taking information about how people are trying to hack into your systems it's not going to use that to train the broader model but it will use that to train your model so it's going to be able to shut down specific types of hacks that might be common in your industry or in your company so it's pretty much like microsoft's bing search engine essentially a cyber security worker is going to be able to type a prompt into a text box and the security copilot is going to fire back a reply based on the app's available knowledge set from kind of its broader algorithm or its broader broader data set so just like the all the other generative AI things that Microsoft is doing they said that their security copilot might sometimes provide incorrect answers to prompts and give people they're going to give people a way that they can just like report that so it's kind of continually going to be able to be trained and improved and of course these things aren't going to be perfect when they launch but as we've seen with chat GPT-3, 3.5 and now GPT-4 these things do get better and they are quite impressive with all of the things that they're able to do.

6:16So this is pretty amazing. Microsoft obviously right now is absolutely riding high on its multi-billion dollar investment in OpenAI. You can see they're trying to take to the full advantage in every way they can and really make it the most profitable possible. It's getting a lot of skepticism and criticism too from some people in the software industry that say, you know, like, look, you have this massive AI model that can do a lot of things. should you be the one integrating it into all the tools but you know it's a free market and so they of course put the money in they can do that so it'll be interesting to see how this continues to evolve in the cyber security field if microsoft becomes a very dominant leader considering they're the first people to launch such a tool and so this is gonna be something we're definitely watching and you know when people are talking about the cons of ai and how it can be used for hacking and whatnot i think at this point the pros are going to be significantly outweighing the cons in a sense that openai is actively working to shut down any you know illicit or bad uses of chat gpt and so integrating it directly into these cyber security offerings and seeing all the good that it's able to do there is going to have a lot of i think net overall positive impact from this technology