Darwinium Launches AI Agent Fraud Tools
Darwinium released two new features to detect AI agent fraud by monitoring journey step changes and MCP tool calls, targeting undeclared agentic transactions.
ImportanceLocalEvidenceE2 unreplicatedWrite-upQuick
Darwinium launched two "intent intelligence" capabilities on October 8 to detect anomalous behavior by AI agents in commercial transactions.
The new Journey Transition Probability feature monitors changes in the order and timing of steps taken by users—whether human, bot, or AI agent—during logins, account changes, or payments. The second feature, MCP Protection, watches the tools agents use via the Model Context Protocol.
The company notes that only about one in four agentic transactions on its network self-declares, and purchases involving agents are rejected nine times as often as others. This update builds on an Agent Intent Detection product launched in March, folding agent tool calls into the same journey view as web and mobile activity for unified risk assessment.