Beelzebub has closed a €3 million seed round, led exclusively by Italian VC United Ventures, to accelerate deployment of its AI-powered cybersecurity platform focused on actively detecting attackers who are already inside networks. The funding will expand the research team, open commercial offices in Rome and San Francisco by year-end, and support customer acquisition across Europe—targeting in particular organizations subject to the EU’s NIS2 directive.
Assumed-breach posture
Beelzebub starts from an “assumed breach” premise: defense is not limited to perimeter controls but presumes an adversary is already present, with the aim of exposing and automatically neutralizing that actor.
Automated offense–defense loop
The company combines offensive and defensive capabilities within a single purple-team structure. Arcangelo acts as the red team, simulating targeted attacks, while Beelzebub Managed provides the blue-team function by deploying LLM-powered decoys—fake servers, databases, APIs and cloud infrastructure that mimic real targets. When an attacker interacts with one of these traps, the platform detects the intrusion, blocks the attack, isolates the impacted host, and triggers automated response actions. The decoys update in real time to mirror newly disclosed CVEs, a mechanism Beelzebub says forces dormant attackers to reveal themselves.
Embedded AI for analysis and reverse engineering
After capture and containment, Caronte—the platform’s AI analyst—performs malware reverse engineering and generates a full incident report “in a fraction of the time” of a human analyst, according to the company. Caronte can also run entirely on-premises to meet the confidentiality requirements of enterprises and government entities unwilling to expose sensitive data to cloud services.
Flexible, compliance-oriented offering
The solution is available as SaaS or as an on-premises deployment for highly sensitive environments, with the option to run LLM inference on hardware supplied by Beelzebub. The company says it is “NIS2-ready by design” and is specifically targeting operators covered by the new EU regulation. More than 60 independent researchers contribute threat intelligence in real time, enabling rapid adaptation to emerging attack techniques.
Funding and financial trajectory
The €3 million seed follows a €300k pre-seed from strategic investors and advisors, bringing total funding to €3.3 million. Beelzebub says it has already begun developing technologies designed to directly protect AI agents.
Context and scope
The startup highlights the urgency of the problem: losses from cybercrime continue to rise, citing figures that put the cost to U.S. organizations at roughly $70 billion in 2025—driven largely by ransomware and cryptojacking—while European rules such as NIS2 and the Cyber Resilience Act impose new obligations and penalties for noncompliance.
Founders and investor perspective
The founders stress the need for defenses that match attackers increasingly enhanced by AI. “Cybersecurity is a non-stop battle that humans can no longer fight alone… Beelzebub adapts to new malware types and is constantly updated to match the state of the most sophisticated attacks,” said Mario Candela, CEO and founder. Massimiliano Magrini of United Ventures added that “the next chapter of cybersecurity will be written by companies like this one,” citing the founding team’s deep technical expertise and execution.
Note: the source text ends abruptly. The original statement begins “Notably, Beelzebub claims to have been the first platform…” but the remainder of that sentence was not provided.
