Blockchain
Bunq (BNK) uses generative artificial intelligence to fight financial fraud
European neobank Bunq is leveraging generative artificial intelligence to combat financial fraud, significantly improving the efficiency of its fraud detection systems. Using NVIDIA’s accelerated computing technology, Bunq improved model training speed by nearly 100 times, according to NVIDIA Blog.
Faster fraud detection
Bunq, which boasts over 12 million customers and holds deposits totaling 8 billion euros, has implemented a highly efficient transaction monitoring system based on artificial intelligence. This system, supported by NVIDIA GPUs, accelerates data processing pipelines by more than five times compared to traditional methods.
Ali el Hassouni, head of data and artificial intelligence at Bunq, highlighted the importance of artificial intelligence in combating financial fraud, a persistent problem in digital banking. Unlike traditional rule-based systems that rely on manually set criteria and generate high false positive rates, Bunq’s AI system uses supervised and unsupervised learning. This approach not only automates the process but also makes it easily scalable.
By using the open source NVIDIA RAPIDS suite of GPU-accelerated data science libraries, Bunq achieved a dramatic increase in the speed of training its fraud detection models. “We chose NVIDIA’s advanced, GPU-optimized software because it allows us to use larger datasets and accelerate the training of new models, sometimes by an order of magnitude, resulting in improved model accuracy and reduced false positives,” el Hassouni said.
Artificial intelligence across the bank
Bunq is extending AI capabilities across its operations. The bank’s proprietary large language model and generative AI power Finn, an AI personal assistant available to Bunq customers. Finn can answer user questions about spending and transaction history, improving customer service.
Additionally, Bunq is exploring NVIDIA NeMo Retriever to further improve Finn’s accuracy. NeMo Retriever, part of the NVIDIA NIM inference microservices, provides models as optimized containers, available with NVIDIA AI Enterprise. “Our initial tests of NeMo Retriever incorporating NIM have been extremely positive, and our collaboration with NVIDIA on LLM is poised to help us take Finn to the next level and improve the customer experience,” added el Hassouni.
In addition to fraud detection and customer service, AI also supports Bunq’s marketing efforts by analyzing consumer engagement metrics to inform future campaigns.
Bunq’s commitment to using AI technology underlines its mission to create a borderless banking experience, always putting the user first. As the digital banking landscape evolves, Bunq’s integration of advanced AI technologies positions it at the forefront of innovation in the financial services industry.
Image source: Shutterstock
. . .