Introduction
Artificial intelligence is bringing innovation in every sector, including cybersecurity. For the best and for the worst. Today we we’ll take the time to discuss where we have seen the most changes in the computer security industry in order to understand the challenges and opportunities ahead of us.
The emergence of AI detection tooling
Intelligent tools are appearing in mass and are helping security experts in all their tasks. Some tools have grown into independent and autonomous detection agents, others serve as a productivity or creativity booster in research.
Jason Haddix, is very popular and successful security researcher, talked about this exact topic as part of his speech of 2025 at DEF CON 32. AI is being used at large scale to monitor malicious traffic and build threat intel which allows for real-time evolution of security scans. This accelerates threat detection and remediation at rates never seen before. As mentioned by Haddix, this also has some controversial sides as ethical security researcher (such as bug bounty hunters) have their attack attempts stolen and used to remediate issues without an opportunity to reward the ethical hacker as it was the case previously. On the flip side, real threats are caught faster and attacks can be prevented at the speed of light without needing human manual analysis which can critically delay the implementation of protective mechanisms. This example highlights great opportunities for defensive detection tooling but also real challenges for industries like bug bounty which incentivizes talented researchers to help companies secure their software.
Security Research assisted by AI
Large Language Models can serve as great tools to help researchers identify weaknesses in software when used wisely. However as we’ll see later this also brings issues.
At selmelc cybersecurity we’ve embraced AI as a creativity and productivity accelerator. LLMs are an incredible tool to gather learning resources and opportunities to continue growing as researchers. From book recommendations to learning about new technologies AI can be a real educational assistant pointing researchers to great resources. It is also extremely effective at scripting which eliminates a lot of unnecessary mental load on repetitive and trivial tasks allowing researchers to focus on the core issues.
However an issue is emerging with this new technology particularly in the bug bounty industry. As documented by Daniel Stenberg (maintainer of CURL) in his blog post . Many cybersecurity enthusiasts are using AI as a security scanner tool, which it is particularly bad in its current widespread form. This leads to software maintainers being swarmed with false positives far worst than in the previous eras of automated security tools. This pollutes responsible disclosure as legitimate reports are drowned in a sea of false positive and taking longer to reach the hands of those that can implement security fixes. This situation currently seems out of control and only getting worst, Daniel has written multiple articles and publicly talked about it numerous times now.
Thankfully as mentioned earlier when used correctly and at the right place AI can have a considerably positive outcome on the researcher’s work. One other example is reporting tools, this is a real pain point for many security researchers as it represent a large chunk of the work. So while it can’t be fully automated just like the research part, any tools that can serve as a productivity boost is positive! Recently we’ve tested Milou.sh which is another security brand from Belgium and offers an AI powered tool which assists in the reporting phase. Such innovations are very welcome, as they reshape our workflows for the best. It has been interesting to see how the research phase and the reporting phase could be linked closer together through intelligent tooling accelerating and at times even improving the quality of reports.
We’re only at the beginning of this evolution of workflows and at selmelc cybersecurity we will be following the evolution of all the different solutions and ideas that emerge from AI.