Artificial Intelligence Act 2021:
The European Commission has been at the forefront of data regulation in recent years, most notably with the implementation of GDPR. Draft regulations have recently been proposed around Artificial Intelligence (AI). The Artificial Intelligence Act drafted on 21st April 2021 addresses the risks stemming from the various uses of AI systems and aims to promote innovation in AI. The Act aims to provide AI developers and users with a clear set of requirements and obligations regarding the use of AI. The draft regulation includes an extraterritorial scope – it extends to systems in countries outside the European Union (EU) where the output enters the EU.
The draft regulations are in the early stages of development. They will be subject to enhanced scrutiny from the European Parliament and member states before finalization and an expected two-year implementation period. Despite its rapid worldwide growth, a critical challenge with implementing the regulations is the lack of global precedent concerning artificial intelligence.
What is Artificial intelligence?
In its simplest form, AI is the development of computers or machines that can perform tasks that usually require human intelligence. Typically, AI is implemented to perform tasks that are difficult for people – or even beyond human capability – by combining computer science and large data sets. Some examples of AI include voice recognition (“Hey Siri!”), pattern recognition, face recognition, machine vision (e.g., Tesla’s assisted driving), and language translation.
Considerations for Banking Organizations
Artificial intelligence has been rapidly increasing in the financial sector in recent years. An increasing number of banks are already using machine learning, and there is a continued motivation to provide automated solutions of existing cognitive and manual routine tasks. This is likely to be common practice as the years progress. AI and automation can process and consume large volumes of data quickly, allowing for more efficient and faster decisions. There are further benefits it brings to the table, such as cost-effectiveness and reducing human error. As time progresses, banks must develop new AI solutions and be aware of the ever-changing landscape to remain competitive.
The financial services industry is highly regulated, which presents a challenge for banks specifically around implementing AI technologies, such as security, ethical, performance, economic, and ultimately reputational risk. To address this, banks must implement a robust framework, including appropriate policies, procedures, training, and controls.
These challenges aren’t new in banking. The introduction of algorithmic trading, an earlier and less advanced form of machine intelligence, brought many of the same issues to the fore. Robotic trading caused many issues – some more public than others – and ultimately, controls around it evolved to include specified business owners, robust processes around updates and testing, “kill switches” to shut down machines going wild, and more. Smart regulators will apply these lessons to AI in banks.
Regulatory compliance is an area where banking organisations have been increasing the use of AI technologies. Navigating the changing regulatory landscape is a challenging process. AI can assist in analysing and consolidating data sources, providing greater insight and understanding than human-driven activities in areas such as contract analysis. It can also support regulatory compliance operations, transaction monitoring, information logging, and audit-related tasks. The potential benefits of AI are stark and can dramatically reduce the cost of compliance with complex regulations, as noted recently by one global bank, which estimated a save of over 50,000 hours of work.
Risk management is another area where AI could benefit banking institutions. Specifically, from a credit and loans perspective, AI has the potential to be an invaluable tool in improving existing processes through the analysis of a broader range of factors, better-informed decision making, and the absence of the limitations of manual systems. However, this is an area where transparency into AI-assisted decision making is crucial as implicit bias can be introduced unwittingly.
Information security will continue to be a key area associated with the banking sector. Artificial intelligence can increase efficiency and allow banks to analyse malicious patterns to detect fraud and money laundering. A continued emphasis will need to be placed on implementing and maintaining security systems to mitigate the risks posed by phishing campaigns, viruses, and attempts to take advantage of existing AI systems.
As AI develops and is more widely used, banks must manage evolving AI regulatory frameworks, significantly impacting how they can use the technology. As things stand, this continues to be a grey area, with AI regulations still evolving and definitions remaining unclear.
With the evolution of technology, there is a fundamental need for banks to adopt a strategic approach, emphasizing using AI tools systematically across broad business processes and functions. It will be necessary for a strategy to be based on the organisation’s scale, size, complexity, and consideration of what client value they are trying to deliver.
The banking industry is still in the early stages of the implementation of artificial intelligence. There will be rapid changes over the coming years which will drive innovation and present new opportunities. The proposed Artificial Intelligence Act regulations will provide a sound platform in achieving consistency in approach across banking organizations, address risks created explicitly by AI, set clear requirements for AI systems, and propose enforcement after AI systems are placed in the market.