Darwinium launched two capabilities aimed at fraud by AI agents. Journey Transition Probability scores each step toward a login, account change or payment against the normal order and timing for that customer, that business and that journey type, flagging steps that look normal alone but suspicious in sequence, whether a person, bot or agent is acting. MCP Protection adds agent tool calls made over the Model Context Protocol to the same journey record as web and mobile activity, verifies agent credentials and can hold payments or other higher risk actions for extra checks, with decisions enforced through the business's existing CDN.
Our readFraud teams preparing for agentic commerce can watch what an agent does across a whole journey rather than judging single events. Darwinium's documentation already showed transition probability features in 2025, so MCP Protection is the newer part, and GA status and pricing are not stated.