FinTechTalk: Outpaced by fraud? Rethinking risk and identity in the age of AI 

Please register or log-in to watch this on-demand episode 

Views on news 

Experian today announced Experian Agent Trust™, a first-of-its-kind framework that establishes a secure, verifiable link between consumers and AI agents, bringing identity and accountability to AI-driven transactions. Without a verified connection between humans and AI agents, autonomous commerce introduces new risks in fraud, misrepresentation and unauthorized transactions.

 

Experian Agent Trust addresses this challenge through a new “Know Your Agent” (KYA) framework, extending identity verification into the age of AI. The framework ensures that agent-initiated transactions are grounded in verified consumer identity. The solution is coupling the user profile with the AI agent through a tokenization process without affecting the customer experience. The new technology, which is also instrumental for merchants who want to embark on an agentic AI journey, is to be released shortly.  

 

How agentic AI is changing the internet 

With agentic AI, a lot of internet features, such as device ID and websites, will become redundant. It’s also expected to change the whole commercial model of the internet. Agentic AI will probably be the most revolutionary new technology since the rise of the internet. Fraud detection must use new methodologies too, as the behavioural biometrics of bots, which are about to become the norm, are currently an anomaly that must be eliminated. So much so that the current setup of e-commerce check-outs, in fact, creates a barrier to the adoption of agentic AI. E-commerce platforms like Seat Geek must have the right controls in place to detect agentic AI-driven types of fraud, even if they don’t pose a problem yet. Email usage is key to gleaning behavioural insights and detecting when a new email without a track record is about to take out loans or other financial products. With bots, scale at speed is always a give-away.  

 

Online vendors must always protect what is critical to their operation and find where the point of monetisation is. Users are usually highly consistent, although they can change considerably when an individual moves house or other major changes happen in their lives. However, a high percentage of rotation is often the indicator of attackers continuously switching the IP addresses, servers or devices used during a login attempt to bypass rate-limits. To help link user behaviour to a person, AtData has a product called NeuroID, which creates a user “fingerprint” based on this data. While the internet is transforming thanks to genAI, there will always be permanent aspects of the end-to-end purchasing process which won’t change and can be harnessed for verification, such as email and delivery addresses.  

 

The panel’s advice 

• Collecting and analysing huge amounts of data on user behaviour is key to fraud detection systems. For success, this data should be combined with the right tools and capabilities.  
• Buying solutions from vendors can speed up adoption of new fraud detection technologies. Vendors also have a broader view of the fraud attempts happening at different organisations at the same time.  
• A small amount of friction can be good and something customers will accept for better security.  
• For AtData blogs and case studies, click here.