Whether you are logging in to a website, subscribing to an online service or making a purchase on an e-commerce platform, every time you establish a new connection to an Internet service, you are presenting an opportunity to cyberattackers. Hackers use techniques such as password theft, data leakage and phishing attacks to steal the identity of an individual with malicious intent.
A Radar to Detect Account Fraud
Orange teams have developed IAlerting to protect Orange customers and users from the consequences of such acts. This system for detecting compromised accounts uses artificial intelligence (AI) to process billions of authentication events, detecting connection anomalies and, therefore, potential cyberattacks. The objective is to evaluate connection legitimacy upstream in order to make the customer experience more secure. “With each authentication attempt, the system analyzes the identification data and assesses whether it needs to notify the customer of suspicious or abnormal behavior,” explains the project team*. “IAlerting calculates the risk score based on statistical modeling of previous connections.”
An Adaptive and Responsive Tool
The system is the result of extensive work to centralize the management of connection sessions. Using centralized data such as the date, time and place of the connection and the device and login used—this data being a mixture of legitimate and illegitimate connections—the project team ran machine learning and AI algorithms to spot suspicious or abnormal behavior patterns. With 100 events to score per second, 20 million customers and 42 million users, the technical challenges were substantial. In spite of this, everything went very smoothly, with the project shifting from the research laboratory to the field in just three weeks! This rapid execution was the result of a user-focused pragmatic approach and collaboration between research and development teams from the outset. To counter the threat from increasingly inventive hackers, the teams opted for a reactive system rather than hard (embedded) coding. “Our real-time scoring tool is flexible and efficient,” the team comments proudly. “It means our researchers can analyze and search in real time for fraud and ways to circumvent it, and our developers can work with actual data and volumes from the development phase.”
A Secure Customer Journey
The booth at the Research and Innovation Exhibition features a demonstration simulating three scenarios based on converting a physical SIM card to an embedded eSIM: a first user who converts their SIM without any problem; a second whose behavior is flagged as suspicious by IAlerting; and a third user, a hacker, who is denied access to the SIM conversion functionality. “In the second example, the user is connected to their company Wi-Fi,” the team explains. “They enter their password but historical analysis reveals suspicious logins, indicating that it is not always them. We then ask for further proof of identity — in this specific case, to log in with the SIM card from their phone. Unless the user’s mobile phone has been stolen, this will confirm that it is actually them!”
Toward Broader Deployment and Greater Effectiveness
Already rolled out in France to professionals and the general public (specifically for accessing the orange.fr site), IAlerting should gradually become more widespread. “Deployment in a new country or on new services is very fast: operational teams can be supplied with a complete platform, ready to integrate local data, within an hour.” On the corporate customer side, Orange Cyberdefense is preparing to start an experiment. “Today, we protect someone who connects behind a web app or a mobile phone,” the team explains. In the future, with Orange Cyberdefense, we will be able to protect a network infrastructure. For example, making sure streaming platform customers do not have their accounts hacked.” Alongside this broader deployment, the Orange Innovation IAlerting team continues to add new features to make the solution more effective. The fight against online piracy never ends!
*The IAlerting project team includes: Erwan Diverrez, Project Owner and Solution Architect for the IAlerting project; Benoit Hérard, Research Engineer for Digital Identity; Christophe Naudin, Head of B2B Identity Enablers providing authentication and access protection to Carrier or Corporate Customer Areas; Maxime Petesch, Orange’s Director of Consumer Security Privacy.