Written by Andrius Palionis, Head of Sales at Oxylabs.
Artificial intelligence (AI) constantly reshapes the cybersecurity battlefield on both sides of the barricades. Cybercriminals use the ever-evolving technology in their attacks on individuals and organizations. Meanwhile, cybersecurity experts utilize AI tools as part of improved network security systems. As AI develops, new risks loom around the corner, and even the best defensive mechanisms are not always enough. On the bright side, our capabilities to proactively counterattack criminals are also developing, allowing us to fight fire with fire. Two technologies are particularly important here — agentic AI and proxies.
The Rise of the Agentic AI
AI, in its various forms, has had a prominent role in cybersecurity for quite a while now. At the moment, agentic AI is gradually taking center stage, especially in organizational settings. Agentic AI describes systems that can act on their own without constant prompting and with limited human supervision.
Thus, they have a kind of agency and proactiveness. Based on pre-programmed plans and objectives, they can make choices to optimize the intended results in concrete situations, not just react to prompts every step of the way.
AI agents are set to create a labor revolution as they can be programmed for various tasks from manufacturing to customer service. Unfortunately, they can also be programmed to conduct various fraudulent activities, such as advanced social engineering attacks that utilize social media data and deepfakes for highly personalized phishing schemes. Thus, Gartner warns that AI agents will accelerate the time it takes to take over exposed accounts by 50% within the next two years.
On the bright side, modern AI tools can also become agents for good. In the UK, Daisy, a chatbot created to act as a perfect phone fraud victim, wastes scammers’ time by keeping them on the phone. This unconventional approach shows that even when we cannot catch the threat actors, we can curtail their ability to hurt others by simply directing their time and energy to futile causes.
We can easily imagine something like that adapted to organizational settings to waste the time of more sophisticated attackers. Additionally, agentic AI systems can be built to analyze real-time threats on their own, block suspicious activity, and even research the cyber threat landscape to make recommendations for future improvements in network security. The possibilities are nearly limitless, and to understand them better, we need to consider another important technology in organizational security.
Proxies and the Time-Tested Measures
Proxy servers are intermediaries between the device making requests and the responding server. When receiving requests, they provide an additional layer where the request can be rerouted and verified before allowing it to reach its intended destination. When sending requests, a proxy provides you with a different IP address so the server does not get to know your real one.
Just like AI, but earlier, proxies were used for both good and bad in cybersecurity. In the latter case, for example, proxy networks can be used as botnets for scams and DDoS attacks. On the other hand, proxies help defend against such attacks and make various cybersecurity measures function more effectively. For example, in sandboxing, proxies direct suspicious emails to sandbox servers where they can be inspected without ever having contact with the internal network.
They also turn the online anonymity, usually enjoyed by the threat actors, against them. Cybersecurity specialists can stay anonymous when collecting open-source intelligence from otherwise hard-to-access online hacker forums using proxies. A network of trustworthy proxies can help collect data on a large scale, especially when paired with AI solutions.
Combining the Forces
AI and proxies, even separately, are both strong weapons that can be used by both sides of the cybersecurity battle. Their combined forces can give a strong advantage to one side over the other, making it crucial for the cybersecurity community to utilize existing and develop new ways to leverage them.
The existing solutions include the aforementioned web monitoring technologies that can detect vulnerabilities and identify emerging threats. AI-powered and proxy-supported infrastructures enable the gathering of such data on a large scale without being blocked by hackers who administer various forums, blogs, and marketplaces where crucial threat data can be found. While doing this with proxies, cybersecurity firms and specialists do not expose their own information. Meanwhile, AI helps to circumvent the measures hackers put in place to ban such web scraping activity.
These innovations also have particular relevance in the crypto and blockchain ecosystem, where privacy, automation, and secure access to decentralized systems are paramount. Blockchain projects, DeFi platforms, and Web3 communities often rely on proxies and AI-driven monitoring to guard against phishing attempts, Sybil attacks, and data scraping by malicious actors.
Proxies also combine forces with AI to scan billions of URLs daily to identify various threats. For example, drive-by downloads, when a user downloads malware simply by visiting an infected website, can quickly compromise your network and allow hackers to steal valuable data. Instead, when AI-powered tools access these websites via a proxy server, they can find the threat without downloading it onto any devices and block its contact with the company network ahead of time.
This is already possible now. The development of agentic AI systems promises even greater possibilities for proxy-supported threat detection and network defense. AI agents created specifically for the purpose with vast proxy infrastructures at their disposal will be able to scan the clearnet and the web on their own, determining the best approaches and self-optimizing to always be ahead of the threat actors. With minimal human oversight, such agents will be able to manage many interactions with scammers, learning from them and adapting to always check any suspicious links, emails, and other online attack vectors before they even get close to their targets.
Like sweet old Daisy, whom scammers call just to waste their time, AI agents can neutralize social engineering attacks by removing a crucial link — human interaction. How else will these agents be able to protect organizations from cyber threats is a matter of innovation by cyber tool developers. However, clearly, investing in proxy-powered agentic infrastructures is the best path forward for cybersecurity firms and the enterprises they aim to protect.
Summing Up
While cybersecurity experts understandably warn about the dangers of agentic AI, there is no reason to give the technology up to the threat actors before it is even fully developed. We already have AI-powered tools for cybersecurity and intelligence gathering. Additionally, our advantage over criminals is that we can do things out in the open. We can build and use vast legally-sourced proxy infrastructures without having to hide in the shadows because we do it for the right reasons.
Thus, the improvement of technology should lead to caution but not panic. And at the same time, it should excite us. With a “whatever they can do, we can do better” attitude, we are well-positioned to use the emerging AI agents and well-established proxy technology to our advantage.
