Understanding the Impact of Gen AI on the BFSI Industry

Amarjeet Khanuja, CISO, Star Health and Allied Insurance

  • No Image

Amarjeet engaged in a conversation with the CIOTechOutlook magazine in order to answer queries on the impact of GEN AI integration into the BFSI industry. He is working as Chief Information Security Officer for Star Health Insurance and has background in technology consulting. He has worked outside India for a decade and has experience in the aviation, retail and oil and gas sector. He has deep expertise in ERP and data governance which he has gained by serving different organizations at different capacities. 

In your view, how can the transformative influence of Generative AI reshape the digital landscape within the BFSI industry segment?

Gen AI is making a lot of inroads not only in BFSI but all across the business ecosystem. The BFSI industry particularly is undergoing a great transformation because of this technology. Traditional AI works on the basis of pattern recognition but Gen AI is multi model. It can create code, images that have the ability to transform customer experiences and streamline business operations. Gen AI is helping to provide personalized customer journeys e.g an AI chatbot can hold engaging conversations with potential customers by offering personalized financial advice and tailoring loan applications as well. In the space of risk management, Gen AI can analyze vast data sets in real time, detect anomalies and help in preventing fraudulent transactions. It can also streamline the operations for all the repetitive tasks like report generation, data entry and others. Gen AI has the capability to improve the operational efficiency using AI powered automation systems. Companies are bale to delve deeper into the analytics of real time data and take informed decisions based on actionable intelligence. This technology can also open the doors for new tailor made financial products and services catering to diverse need of customers.

AI model bias can lead to unfair treatment of certain customer segments. What strategies would be employed to manage the risks of AI model bias in BFSI applications?

Decisions taken by AI models are based on the data sets that are being used to train them. Companies need to be very careful about data transparency and scrutiny. Here, it needs to be understood that unbiased data is crucial for accurate performance of AI systems. Understanding the origin and composition of the data is essential for training AI models. Regular audits and unbiased detection techniques need to be followed in order to achieve the desired outcomes from AI systems. There also needs to be diversity in the development teams to ensure all round development. Observability is one of the key requirements in cyber security today and also for network operations and software operations. For AI systems, it is called explainability. We need to assess the factors to drive a successful AI models and for this proper human oversight is necessary. Here, BFSI organizations have to continuously monitor their AI models and continuously train them by means of diverse data sets to achieve the desired outcomes.

With the increasing use of Generative AI, How do businesses address potential data breaches that could arise from Generative AI models processing sensitive information?

The industry can provide ample examples in this regard. One of the notable ones was detected in Samsung where one of their codes were uploaded in ChatGPT which got exposed. In order to address this problem, organizations need to have strong data governance framework. They need to have clear information about the sensibility of data and how they are stored and which people from the organization have the access to that. From the perspective of information security, less is more. AI models need to be trained using only the required amount of sensitive data. Organizations also have the option of using synthetic data in this regard. In case organizations need to use original data, then anonymity must always be followed along with tokenization techniques to safeguard sensitive information. They should establish proper access controls to check accurate access of data and supervise the overall functioning of the system.   

According to you, what cybersecurity measures are critical when integrating generative AI into BFSI systems?

It primarily depends on how organizations are implementing proactive defense. Threat intelligence must be incorporated in this regard. Organizations need to hire good threat intelligence network and act on the problems that get detected in a time bound manner. Second is monitoring of the attack surface for which regular penetration testing is required. Secure coding practices need to be followed as well for integrating generative AI into BFSI systems. Principles like "security by design" and "privacy by design" are much relevant in this regard.

🍪 Do you like cookies?

We use cookies to ensure you get the best experience on our website. Read more...