In this edited excerpt from a recent Nasdaq TradeTalk, Check Point’s Head of Engineering, U.S East, Mark Ostrowski, discusses the concerns that ChatGPT-4 raises.
ChatGPT-3 has evolved into ChatGPT-4. What does that mean and why is that giving bad actors a dangerous amount of power?
It’s a good question. I think that what we’ve discovered since November of last year, when ChatGPT was first released, is that ChatGPT has really opened the doors to a lot of different things, some of which are positive…and some that we can discuss today.
For example, are these AI bots becoming sophisticated to the point where they’ll start to contribute to elevated levels of cyber crime?
In terms of the evolution from ChatGPT-3 to ChatGPT-4, it really just means that the language processing capacity is getting better; that it’s a more sophisticated tool, that things have been fine-tuned to a higher level, again opening doors to threat actors leveraging the tool even more than previously.
ChatGPT has user safeguards built in. How can those restrictions be disabled?
Based on the safeguards in-place, if threat actors (or even just regular consumers) identify words that the tool is programmed not to respond to, they can effectively navigate around the safeguards by changing the nature of their requests.
And when threat actors do that, they can improve phishing templates to make them more believable…
How concerned are you about the security of something that you cannot oversee? Whether that’s the iteration of ChatGPT-4 or future versions?
These are valid concerns…From a cyber perspective, I think it’s pretty simple. There are a lot of known attacks for which the other side of the house – the cyber security professionals – have to create better prevention. That has to be the first step.
ChatGPT isn’t really creating a lot of these outcomes. It’s leveraging things that are already out there. So implementing the safeguards that the cyber security community already has in place will create a certain level of prevention, and then we can work on how to move forward and how to prevent more sophisticated threats related to AI-bots.
Is the cyber security industry prepared to deal with the era of ChatGPT and AI?
When you think about cyber security, remember that machine learning and artificial intelligence have been there for many years…Artificial intelligence and machine learning are foundational in enabling us to prevent threats.
Now we’re seeing the threat actors leveraging the same tools. So, when you think about, as you mentioned, that next evolution of cyber security, what you’re really talking about is almost AI vs. AI in some ways. These tools can be leveraged to create that dynamic.
To give a sports analogy, what’s the best offense? A really good defense. For any person or company that’s looking to create better preventative cyber security outcomes, ensure that artificial intelligence and machine learning are part of programs, projects and products, because you need that in order to combat what’s out there today.
Watch the full clip here. Interested in learning more about ChatGPT, artificial intelligence, leading through change and whether or not AI will result in job cuts? Get leadership insights for the AI-optimized future, here.