Cyberfraud prevention firm Darwinium has launched Beagle and Copilot, two agentic AI features that simulate adversarial attacks, hidden vulnerabilities, and optimise fraud defences.
As AI evolves, it can be utilised by fraudsters to evade detection and manipulate digital systems. Darwinium aims to deliver defenders their own form of AI built natively into its behavioural intelligence platform and allowing fraud teams to spot anomalies and separate good agents from malicious ones in real time. The company believes that the main challenge faced by fraud detection teams is distinguishing AI agents acting on behalf of good users and malicious automation.
As detailed by Darwinium's CEO and co-founder, Alisdair Faulkner, "agentic AI is emerging as a transformative sales channel, reshaping an eCommerce ecosystem originally designed for human interactions. This shift presents a new challenge: distinguishing between genuine users, fraudsters, and autonomous AI agents. As AI-facilitated fraud grows in sophistication and scale, businesses must counter it with equally advanced defences. In other words, to fight AI, we need better AI. And the foundation of better AI is better data. Organisations need continuous visibility into behavioural signals across the full customer journey, capturing every digital interaction to build a high-fidelity picture of intent."
Tools to simulate and detect malicious agents
Beagle is an agentic AI designed to identify blind spots in fraud protection by simulating adversarial behaviours and enabling teams to test detection and mitigation strategies in production-like environments. The tests emulate synthetic identities by generating realistic user profiles, and identify weak links in defences such as device fingerprinting, behavioural biometrics, and bot scoring systems. Beagle navigates complete user journeys, from account creation to checkout and support interactions, simulating attacks like false identity creation using disposable emails and virtual numbers, credential stuffing, behavioural mimicry, residential proxies, and captcha solving.
Beagle integrates into Darwinium’s architecture to offer real-time analysis on system response, fraud detection capabilities, and unexpected failure models. It feeds this information to the platform’s decision layer to improve risk policies, adapt detection logic, and upgrade mitigation strategies without manual coordination between fraud, product, and engineering teams.
Copilot, the AI-driven companion designed to simplify risk engineering, acts as an intelligent assistant for fraud remediation strategy, decision-making, and platform interaction. It focuses on simplifying user queries around complex fraud detection and, once suspicious activity is found, it makes fraud investigation recommendations and suggests remediation strategies.
For more information about Darwinium, please check out their detailed profile in our dedicated, industry-specific Company Database.
