Gl00be introduces a AI-based, Prague-born Cybersecurity Tool
Much like drones have done to warfare, AI has transformed the field of cybersecurity.
“Everything now is faster,” Diego Martinez, Gl00Be’s Chief AI and Security officer, says. “And it is not a matter of few degrees. This is a complete change in scale and speed.”
The change came when large language models came online. These models offered hackers an extraordinary new tool to seek and exploit vulnerabilities in software. Their ability to attack much faster, at a wider scale, and with the ability to alter their angle of attack based on outcomes increased their advantage of surprise.
“Hackers armed with AI have so much more capability to bypass current security systems,” Radwan Seddiqi, Gl00be’s CEO, says. “It is a little like taking candy from a baby.”
This new capability particularly threatens open source software. All software is as vulnerable theoretically. Being vulnerable is a function of the quality of the software design and coding. However, proprietary software is backed by a company with the interest and resources to fix vulnerabilities fast and fully. Open source software depends on a community of software developers who may have the desire to fix vulnerabilities, but not the resources.
“Open source software faces a big challenge,” Martinez says. “It is not the only challenge, however.”
Equally daunting are the legacy systems on which many major institutions entire digital infrastructure are based.
“These systems were designed a long time ago,” Martinez says. ‘When they coded these systems, they were developed just to work…to do the task the company needed. No one conceived that they would be connected to the world through the internet, or that an entire complicated system of software would be built on top of them, or that a whole world of criminal enterprises or state security operations would be born specifically to cripple the software. The people who wrote that code are retired now, and what they know about how that software works may no longer be known by anyone currently working in the company. This creates a massive vulnerability sitting deep down in the heart of system.”
This analysis has recently been verified by the introduction of Anthropic’s Mythos model. This model has proven exceptional at detecting vulnerabilities in software systems.
“Mythos is a game changer, but it is not only one out there,” Martinez says. “In a recent test, it clearly outperformed all the others. ChatGPT’s model also increased its ability by significantly, and did better than Mythos in detecting kernel vulnerabilities, which are the most serious type of vulnerabilities.”
So, once the blood starts flowing again through the executive mind, what can companies do to react?
“Companies who wait will have it the worst,” Martinez says. “We need to attack, and we need to attack now.”
The first step is integrating AI as much as possible into their cybersecurity.
“Cyberattackers are always going to maintain some sort of “first-mover” advantage, because they can devise a new way to use the tools they have that cyberdefenders have not considered fully,” Martinez says. “It is going to be a constant act-and-react environment. If your company is not fully utilizing AI, you are not going to be able to react quickly and forcefully enough.”
That means companies will likely need a suite of cybersecurity tools that allows them to react to different attacks with different defenses.
“The first step is to detect your own vulnerabilities and then fix them,” Martinez says. “You need to attach your system from every conceivable angle, and develop a list of what is broken and how to fix it.”
That list is important, because companies are likely to lack the manpower or financial resources to fix everything at once, particularly companies highly dependent on legacy systems.
“You are going to have be a little like the manager of an emergency room in a major hospital,” Martinez says. “You are going to have to assess risk levels and resources required, and prioritize what needs to be done now and what will have to be left until later.”
Gl00be is introducing a new AI-Cyber Defense Engine that can assist with that process. Gl00be says it is the 13th application in its area that Microsoft’s Azure has certified to run on its platform.
“In this new world of AI, companies are going to have real challenge to defend themselves, and do so economically,” Seddiqi says. “Many large enterprises can pay up to USD 500,000 annually on cybersecurity. Our approach is to offer a license for USD 389 for use of digital tool.
“Our system can orchestrate how your software system detects vulnerabilities and attacks,” Martinez says. “And we are working to add AI-supported tools for fixing vulnerabilities quickly.”
Gl00be’s vision ranges out beyond the borders of software into robotics. They recently signed a contract in Mexico to combine software with “robotic guards” to control the opening and closing of prison doors. They hope to use that experience for other projects that co-ordinate software and hardware.