Apply at
Request a demo
Be called back
Crédit Agricole Payment Services and AdvanThink are working together to develop sustainable Artificial Intelligence to enhance confidence in payments.
Summit for Action on AI - On February 10 and 11, Paris hosted the Summit for Action on Artificial Intelligence, an international event organized by the Elysée Palace, during which the entire AI ecosystem came together to build the future of this technology and its uses.
On this occasion, Crédit Agricole Payment Services and AdvanThink jointly presented their payment security project as part of the "AI for efficiency" call for expressions of interest issued by the Direction Générale des Entreprises. The project, entitled "Protect payments in real-time aims to use real-time Machine Learning algorithms to identify and prevent bank payment fraud. It has been selected as one of the winners of the Call for Expressions of Interest.
Key points of the "Protect payments in real-time" project:
- Crédit Agricole Payment Services relies on FraudManager, AdvanThink's real-time fraud detection software, to secure its customers' real-time payments.
- Artificial Intelligence models for fraud detection are deployed and scan massively increasing amounts of data, identifying in real time whether a transaction presents a risk of being fraudulent. As fraud techniques evolve, models are updated to guarantee optimal protection, day after day.
- The collaboration between Crédit Agricole Payment Services and AdvanThink enables the realization of a critical use case for the company, while adopting a sustainable approach that limits the impact on resources and the environment.
Securing payments in real time
Payment fraud is increasingly based on the exploitation of human weaknesses. While preventive measures can raise user awareness, they are not enough to contain the evolving practices of fraudsters in terms of social engineering and data theft.
For Crédit Agricole Group banks, it is essential to protect their customers against fraud.Crédit Agricole Payment Services invests heavily to offer innovative, high-performance, secure and robust solutions.
But it is also essential, in a world where everything moves faster, to enable citizens to use their means of payment in a seamless process. In the fight against fraud, we need to protect each individual without them even realizing it, in the space of a few milliseconds, in a process in which fraudsters are using increasingly advanced techniques.
The challenge for Crédit Agricole Payment Services is to ensure that payments are secure at the very moment they are made.
Mass data processing, industrialized and real-time Artificial Intelligence
To detect payment fraud, it is necessary to identify, from all the data available at the time the transaction takes place, whether the current payment presents a risk of being a fraudulent transaction.
In this context of mass data exploitation, identification of common patterns of fraudulent behavior, real-time context and improved prediction, the use of Artificial Intelligence models is totally appropriate.
By thwarting fraud attempts, these Artificial Intelligence models help to consolidate confidence in payment methods.
For real-time bank payment fraud detection, the Artificial Intelligence models in FraudManager, AdvanThink's fraud detection solution, exploit the following data in particular:
- Payment service user information
- Payment information
- Payment context information
Guaranteeing response times of just a few milliseconds, this use of AI makes it possible to protect cardholders from the actions of fraudsters by providing a risk score that quantifies the risk of fraud (card theft/loss, identity theft, phishing through behavioral change, etc.) before the payment is actually authorized.
Invisible" Artificial Intelligence with multiple benefits for Crédit Agricole banks and their customers
Thanks to the models deployed with FraudManager, Crédit Agricole Payment Services proactively detects and blocks several hundred million euros a year in pure financial losses.
At the same time, the industrialization of the system aims to improve the efficiency of anti-fraud teams, by reducing investigation times, automating processes and prioritizing alerts, for example. This enables human resources to be reallocated to higher value-added tasks, such as the analysis of complex fraud patterns and the continuous improvement of AI models.
Finally, AI helps maintain an optimal balance between security and payment fluidity for customers:
- Reduced unjustified blocking of legitimate transactions thanks to improved risk scores.
- Increased customer confidence thanks to enhanced security and greater responsiveness to fraud.
Sustainable Artificial Intelligence for a critical use case
The question was at the heart of the Summit's debates: are industrialized Artificial Intelligence models, based on mass, real-time data, necessarily an energy sink?
The answer, of course, is no! At a time when the growth in data production is exponential and the impact of digital technology on the environment is well known, Crédit Agricole Payment Services has incorporated the challenge of limiting the environmental impact into its AI project.
Since its creation in 1990, AdvanThink has incorporated a "frugal by design" approach into the development of its solutions. The use case performance approach is compatible with a responsible approach, limiting environmental impact, even if the project is critical.
Thus, environmental impact is continuously monitored and limited to 3 levels:
- Optimizing resources: the energy consumption of our infrastructures is constantly monitored, and measures are taken on an ongoing basis to improve performance and resource utilization.
- Efficient model design: AdvanThink prioritizes lightweight, energy-efficient model design, requiring fewer computational resources while maintaining optimum performance.
- Renewable energies: wherever possible, we deploy our Artificial Intelligence solutions on servers and services powered by renewable energies, thus reducing the environmental impact of technology use.
Would you like to find out more about our payment fraud detection use cases?
