Artificial intelligence fraud prevention platform company Darwinium UK Ltd. today announced the launch of a new intent-based authentication and orchestration solution designed to secure the rapidly emerging world of agentic commerce.

As AI agents increasingly browse, purchase and manage accounts on behalf of consumers, businesses face a new security reality. Darwinium argues that because of the rise of legitimate agents, blocking automated traffic outright is no longer viable. But at the same time, attackers are exploiting the same technologies to scrape content, stuff credentials and hoard inventory.

An example of the problem of scraping and hoarding of inventory was detailed in a report by DataDome SAS earlier this month. Surging demand for DDR5 memory, fueled by the AI boom, has triggered a wave of automated scalping activity that is overwhelming online retailers and distorting supply.

Darwinium’s new agent intent intelligence introduces a new model for digital trust that goes beyond simply authenticating the agent. It also evaluates behavioral intent and dynamically applies the appropriate level of friction based on risk.

The solution, which can be deployed natively at the edge across major content delivery network providers such as Cloudflare and AWS CloudFront, determines whether a request originates from a verified AI agent, a human user, or malicious automation. It then analyzes real-time behavioral signals and journey context to decide whether to permit, verify, challenge, or prevent the interaction.

“AI agents are already acting across e-commerce and financial services,” said Chief Executive Alisdair Faulkner. “The real challenge is distinguishing trusted automation from abuse and applying the right level of trust at the right moment. Our intent intelligence gives organizations the intelligence to do exactly that.”

Darwinium’s solution provides a governance layer for delegated digital activity, with every request evaluated in real time and orchestrated through four adaptive responses: permit, verify, challenge and prevent.

Permit allows legitimate, low-risk agent interactions such as browsing and product discovery, governed by velocity and device controls and verify determines when signals shift by enriching them with additional identity or device intelligence. Challenge looks at activity through silent or step-up authentication when financial or account risk emerges. It stops malicious automation from executing high-risk actions such as credential stuffing, scraping or inventory abuse.

Darwinium says that because enforcement occurs at the edge, the controls can be applied at critical moments such as login, account changes and checkout before fraud propagates downstream and without introducing latency.

The approach taken by the new intent-based authentication and orchestration solution is also designed to interoperate with evolving payment and identity protocols that support delegated transactions. They include Visa Trusted Agent Protocol, Mastercard Agent Pay and emerging Know Your Agent models.

Darwinium’s new solution is available immediately across all Darwinium edge deployments and requires no additional infrastructure.

Darwinium is a venture capital-backed startup that has raised $26 million in funding, including a round of $18 million in October 2023. Investors in the company include U.S. Venture Partners LLC, Blackbird Ventures Pty. Ltd., Airtree Ventures Pty. Ltd. and Accomplice LLC.

Image: SiliconANGLE/Ideogram