International Journal of Scientific Research and Engineering Development

The massive shift toward online networks and cloud computing has directly caused a massive increase in digital attacks. Hackers no longer rely on simple tricks or hitting a single target hoping to get lucky. Today, they move step by step through a company’s network, looking for different vulnerabilities to exploit over weeks or even months. Standard protection tools are struggling to keep up because they mostly work completely alone. For example, a network firewall does not talk to the antivirus program running on an employee’s laptop. Because these defensive programs refuse to share data or work as a team, smart attacks easily slip right past them. To solve this massive gap in modern network defense, we designed a completely new security model that relies on multiple artificial intelligence agents working together. This new setup uses a central controller, which we call the Agent Orchestrator. This main boss oversees three very specific workers: one agent dedicated entirely to finding active threats, one agent for jumping in and responding to those incidents the second they happen, and a third agent that quietly works in the background to find and manage system weak spots before hackers do. The main controller collects all the clues from these three workers, decides the best course of action, and updates a shared memory bank to catch large-scale attacks that span the entire network. By tying together the acts of monitoring, fixing, and responding, the model offers a complete, full-circle defense shield. We also added language models to the mix so the system can actually write clear, plain-English explanations for every single alert it generates, making life much easier for the humans running the network. During our testing phases, this design achieved a 94.2% accuracy rate. It also drastically cut down on annoying false alarms and handled real security incidents much faster than older methods. These outcomes heavily prove that using a coordinated, multi-agent layout is a highly practical and effective option for real-world enterprise companies looking to secure their data.

Rakshitha D V, Shilpa B, Sandesh S, Swatantra Deo Swami,"Multi-Agent Threat Detection Using Cybersecurity System" International Journal of Scientific Research and Engineering Development, V9(3): Page(781-788) May-June 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.