AI Threats in Banking and Realistic Security Approaches
An IBM security analysis examines AI-powered fraud and automated attacks in retail banking, discussing current risks and authorization models.
An IBM security analysis examines AI-based threats in the retail banking sector, highlighting the differences between fraud, automated attacks, and future authorized agent models.
The Scale of AI Threats in Banking
The primary risks highlighted in traditional retail banking channels are AI scaling fraud and non-human automation attacking existing systems.
The Difference Between AI and Agent-Based Systems
While AI provides content generation or decision support, agent-based systems have the ability to take action toward a defined goal with limited human intervention.
Developments in E-Commerce and Banking Channels
While agents that shop on behalf of users are actively used in the e-commerce space, a similar model is not yet directly supported in retail banking interfaces.
Open Banking and Authentication
Traditional banking builds third-party access on strong customer authentication and explicit consent principles under Open Banking regimes such as PSD2.
Automation of Fraud Methods
While attackers make known operations such as social engineering, phishing, and fraud faster and more scalable using AI, banks must focus on behavioral intelligence.
Future Authorized Agent Models
With the proliferation of personal agents, it will become difficult to distinguish between legitimate authorization and account takeover situations, making contextual risk assessment critically important.