rewrite this title Darwinium Launches AI Tools to Identify Adversarial Fraud – Finovate

Cyberfraud prevention company Darwinium launched a pair of new agentic AI solutions this week.

The new agents—Beagle and Copilot—enable red-teaming simulations and AI-assisted remediation to help financial companies enhance their fraud defenses.

Headquartered in San Francisco, California, Darwinium made its Finovate debut at FinovateEurope 2023 in London.

San Francisco, California-based security and fraud prevention firm Darwinium unveiled new tools to help organizations deal with the challenge of adversarial attacks, detect hidden vulnerabilities, and enhance overall fraud defense. This week, Darwinium launched a pair of AI agents—Beagle and Copilot—to give companies their own autonomous AI capabilities to combat those AI agents deployed by fraudsters.

“Consumers are using AI agents to shop faster. Fraudsters are using them to bypass defenses at scale,” Darwinium CEO and Co-Founder Alisdair Faulkner said. “The challenge isn’t detecting bots anymore, it’s distinguishing AI agents acting on behalf of good users versus malicious automation. Solving that problem requires a platform that speaks the same language: agentic AI.”

Beagle enables firms to conduct AI agent red-teaming to detect blind spots in their fraud defenses. Beagle simulates complex adversarial behavior, which enables fraud prevention teams to test their detection and mitigation strategies. Beagle generates realistic user profiles, spoofed devices, geolocation variance, and other tactics to emulate synthetic identities. The technology simulates a variety of attacks, including credential stuffing, behavioral mimicry, captcha solving, and more, and integrates directly into Darwinium’s architecture to provide real-time detection and responses.

Darwinium’s Copilot acts as an intelligent assistant that supports an organization’s remediation strategy, decision optimization, and platform interaction. The technology streamlines fraud team queries for different types of fraud detection circumstances such as “users who exhibit bot-like activity or are creating synthetic accounts.” Once these users are identified, Copilot can then conduct fraud investigation recommendations and offer remediation strategies. The solution automates the feedback process from detection to decisioning to business impact, making Copilot especially helpful in high-scale environments.

The new offering from Darwinium comes as fraudsters are becoming increasingly effective at exploiting AI bots for nefarious purposes. Because of this, as Faulkner noted in a recent thought leadership piece for Dark Reading, it is all the more important for security and fraud teams to focus on and discern the intent of AI agents rather than the mere presence of these entities. Legitimate agents, Faulkner explained, tend to act in ways that resemble trusted user behaviors and patterns. AI agents that are malicious, on the other hand, conduct themselves in identifiable ways—such as skipping normal browsing steps to access high-value endpoints (like login and checkout) or generating a sudden high volume of API calls. In Faulkner’s estimation, technologies such as Beagle and Copilot help institutions and their fraud teams move from “What is this AI agent doing?” to “Why is this AI agent doing this?”

Launched in 2021 by the co-founding team that built and scaled ThreatMetrix, Darwinium made its Finovate debut at FinovateEurope 2023. The company raised $18 million in funding in the fall of 2023 in a Series A round led by U.S. Venture Partners.

Photo by elnaz asadi on Unsplash

Views: 360