Decoding the Impact of ChatGPT for Treasury

Generative AI, represented by groundbreaking models like ChatGPT, is poised to revolutionize treasury management. However, its integration brings forth a spectrum of implications, from promising opportunities to potential risks. We highlight the potential benefits and risks associated with integrating generative AI in treasury and finance operations.

Exploring Generative AI

Generative artificial intelligence (AI) is a subset of artificial intelligence technology. Unlike traditional AI systems that are often limited to specific tasks or domains, generative AI models are designed to generate novel outputs, including text, audio, images, and even entire virtual environments.

Put simply, these models are usually trained on large datasets to learn patterns and relationships within the data. This allows generative AI models to create new content that is coherent and relevant to the input provided..

One of the main characteristics of generative AI is its ability to produce diverse and realistic outputs that mimic human creativity and intelligence. Although it has the potential to automate manual tasks and improve efficiency, there are concerns about fake or misleading content and the potential for misuse like deepfakes.

ChatGPT Is a Pioneer in the GenAI Space

A prominent example of generative AI in action is ChatGPT, developed by OpenAI. Since its launch, ChatGPT has garnered widespread acclaim, reaching 180.5 million users by March 2024. OpenAI and ChatGPT gained further notoriety by recently partnering with Microsoft for a multi-year alliance valued at $10 billion.

Google has also taken notice; even though ChatGPT is not a search engine, it is already being viewed as competition for the online search giant. Google’s management has developed a chatbot of its own, though in a recent demo, it made a mistake that has already cost the company $100 billion in market value.

What GenAI Means for Treasury

GenAI, including ChatGPT, offers a range of new options for treasury teams, including the ability to ask questions of your treasury data – typed or by voice – as well as build new reports, queries, dashboards and even new programming and scripts.

The biggest challenge for treasury teams is how to manage GenAI in a secure and safe manner.

Different organizations will have different policies around the use of tools such as ChatGPT, including when open source language models can be used versus closed models such as ChatGPT Enterprise.

To be responsible to all their customers, including those organizations that prohibit the use of ChatGPT altogether for their finance teams, most treasury system providers will not (and arguably should not!) embed ChatGPT directly into their software platforms. Rather they are opening their data models through APIs to make treasury data available to GenAI tools – which allows treasury teams to make their own choices about what, if any, GenAI tools they integrate into their treasury processes.

For those that choose to introduce ChatGPT and similar tools into their treasury operation, here are examples – both good and scary – where we have seen GenAI increase its relevance for treasury:

Payments Fraud Detection

It’s been well documented that AI is being used regularly for both payments fraud and payments fraud prevention. Generative AI has now become the latest tool in fraudsters’ arsenals – making it a ‘scary’ scenario for treasury teams to combat.

Cybercriminals are using generative AI tools–including ChatGPT–to help craft sophisticated phishing messages. There has been a 1,265% increase in malicious phishing emails since Q4 2022. Check Point, another cybersecurity firm, noted that using ChatGPT helped it create an end-to-end social engineering campaign from phishing emails all the way to embedded malware within an email attachment. The result was disturbingly convincing and confirmed that GenAI tools have lowered the bar for code generation, making it easier than ever for bad actors to attempt fraud.

In response, cybersecurity experts all say the same thing: use automated systems to detect and prevent cyberattacks at machine speed. In treasury, this presents an opportunity to use AI in the payment process to detect suspicious payments, using adversarial networks to identify payment anomalies compared to an organization’s own payment history.

While this is an example of GenAI being used for nefarious purposes, embedding a closed AI model to use only an organization’s own payment data to identify potential fraudulent payments is a great way to use AI to fight the AI employed by fraudsters.

Enhancing Excel with AI

Imagine a world where generative AI technologies seamlessly integrate with essential business tools, transforming traditional treasury management systems without disrupting existing structures. This vision is now a reality with Excel, supercharged with AI tools like ChatGPT and Copilot through OpenAI’s partnership with Microsoft. This integration not only enhances workflows with secure and streamlined interactions but also revolutionizes how data is managed and presented.

Excel’s AI-powered features offer a GenAI function to engage with your data, capturing insights from both internal models and external sources. With Copilot, users can effortlessly create complex and easy-to-read spreadsheets. Need an extra column? Copilot will create it. Forgot that unique combination of formulas? Copilot has you covered. ChatGPT’s large language model unifies the data you need, while Copilot formats it into the most impressive spreadsheet you’ve seen.

Microsoft addresses automation and data security concerns with Appsource store’s treasury apps, which automate the secure delivery of data into your spreadsheets. These embedded apps allow data to be queried from the cloud and presented directly into your spreadsheet. Simply sit back and enjoy the AI ride as your treasury management capabilities become more efficient and secure.

Bank Payment Formats with AI

One of the most challenging capabilities for both finance and IT teams is the bank payment format transformation process. This difficulty comes with the multitude of format variations banks require, making a standard such as XML ISO20022 PAIN v9 into more of a guideline with sometimes dozens of different file format requirements, even within the same bank.

ChatGPT becomes a great solution to build these unique file format customizations on the fly for a payments system user or the IT professional tasked with updating the file delivery from their ERP to the bank. The ERP or treasury system exposes, via secure API, the payment script to ChatGPT or a similar GenAI tool, where the user asks the AI to update the script to meet the payment format scenario in question. The resulting script can be applied to a payment file in real-time, or saved and applied to the next payment run.

The value here is empowering the user to make customizations, rather than being tied to a vendor or consultant that will delay your process and charge you for the privilege.

The Future of Generative AI

The introduction of ChatGPT-4 will only serve to expand the adoption of chatbots and large language models for the common user, paving the way for greater adoption within the business community and corporate finance teams.

For treasury system users, ChatGPT and similar GenAI tools will replace mouse-clicking to reach a menu item or open a screen with voice activation. It will not be long before we can ask questions such as “When should I hedge?”, “How do I reduce bank fees?” or “Which banks support SWIFT’s UETR tracker?”

Executives, who do not typically login to TMS or ERP platforms, would no longer have to rely on reports, but could engage with data and insights directly. Even the least tech-savvy CFO can ask a question of ChatGPT.

From the promise ChatGPT has already shown, mass adoption in finance and treasury is around the corner, once CFOs, CISOs, and legal teams have complete confidence that their organization’s data is secure and the results are accurate.