Max Justicz, CTO of the startup Charge Robotics, recently disclosed a critical vulnerability in the widely used open-source Ubuntu operating system. This discovery, made with the help of an advanced artificial intelligence model, could have allowed attackers to take control of servers running the system if exploited.
Justicz, who also dedicates his free time to cybersecurity, stated that he identified the problem using one of the powerful models. He described how he simply pointed the tool at the same code block where he had previously found another bug and asked it to find similarly severe or worse issues. Within just two hours, the system uncovered a vulnerability that is extremely difficult for a human analyst to detect, yet fully functional.
The discovered vulnerability was reported to the Debian project, which manages the relevant software, and a patch was subsequently issued to fix it. Although no exploits are known yet, the ease with which the bug was detected suggests that we are witnessing only the beginning of an era of AI-assisted exploits, which could be leveraged by even less experienced attackers.
Justicz warns that in the near future, we can expect the discovery and exploitation of "extremely insane" bugs, as the effectiveness of pointing a powerful AI model at unaudited code is enormous. It is clear that this represents a significant "alpha" for anyone who decides to explore code or binary files that no one has yet scrutinized.
Major entities developing artificial intelligence have noted this threat and are responding. A new cybersecurity-focused model has been announced for selected users within a program that provides tools for vulnerability discovery and development. This model was reportedly used to scan and prepare patches for vulnerabilities in the Linux kernel, where it identified nearly a thousand potential issues, and discovered dozens of previously unknown vulnerabilities in the Firefox web browser. The goal is to dramatically broaden access to advanced security research for organizations of all sizes, independent researchers, and open-source project maintainers.
There is a belief that the window of opportunity to secure digital infrastructure is limited before AI models become even more effective in offensive scenarios and simultaneously more accessible. For defenders, however, this could mean an enormous amount of new work that exceeds their resources. Justicz himself spent significant funds on subscriptions and access to models to discover several exploits. In addition to access to advanced models within the mentioned programs, substantial financial support is also provided for cybersecurity professionals working on critical infrastructure, such as power grids or water management systems.
The cybersecurity program also faces criticism. For instance, some of its functionalities were reportedly "weakened," limiting its utility for defenders. On one hand, this might be an attempt to restrict access for potential attackers – Justicz mentioned that he joined the program merely by uploading a photo of his driver's license. On the other hand, cybersecurity professionals feel disadvantaged if they cannot utilize cutting-edge technology, while attackers can deploy open models with security restrictions removed.

Need a hand with your website or app?
Get in touch and I will reply within 24 hours with a proposal for the next step.
Although "thousands" of cybersecurity professionals are reportedly involved in the program, this number pales in comparison to the much larger number of potential attackers. The question remains whether sufficient resources are being dedicated to this problem. While significant computational capacity is allocated pro bono for discovering and patching vulnerabilities in open-source code, the costs for such "bug hunters" can be substantial. For example, finding bugs in Linux cost 1.2 million dollars, which is significantly less than the investment to solve a famous fluid dynamics problem. If the time to patch the digital infrastructure on which the world depends is truly limited, then urgent action with greater priority is imperative.
