24Fintech.

com | #24Fintech

Governance of AI
Technologies in the
Financial Services
Report of the Roundable held during
24 Fintech on 3rd September, 2024

Hosted by: Co-organised by:

1
Roundtable Discussion on the
“Governance of AI Technologies in
Financial Services” at 24 Fintech

At the 24Fintech event in Riyadh, CFTE, as a intersection of AI and financial regulation.

knowledge partner, hosted the roundtable While not an exhaustive industry analysis,
titled “Governance of AI Technologies in this report distills key takeaways from
Financial Services,” uniting 12 prominent thought leaders at the forefront of AI
leaders from central banks, regulatory governance. Their expertise sheds light on
bodies, and the financial industry to confront creating adaptable regulatory frameworks
one of today’s most urgent challenges: the that balance the need for innovation with
governance of AI technologies in financial the imperative to protect consumers and
services. As AI continues to reshape the sector, ensure ethical AI use.
the need to effectively regulate and harness
its potential has become a top priority for We extend our sincere gratitude to
policymakers, industry leaders, and regulators the speakers who generously shared
worldwide. This roundtable provided an their expertise and perspectives. Their
essential platform for experts to exchange contributions have significantly enriched
insights, tackle emerging challenges, and the ongoing dialogue on the responsible
explore the future of AI governance. governance of AI in financial services and
will play a crucial role in shaping the future
The discussions focused on three core of AI regulation.
themes, each addressing the complexities
of AI regulation in finance. These themes As you explore this report, we hope it
formed the basis for evaluating current serves as a valuable resource for deepening
frameworks, identifying regulatory gaps, and understanding, fostering collaboration,
addressing the ethical risks associated with AI and guiding efforts to build a sustainable,
deployment in financial systems. globally aligned regulatory environment for
AI technologies.
The insights captured in this report are
indispensable for those navigating the

2
Contributors
Governance of AI Technologies
in the Financial Services

CO-CHAIRS

HUY NGUYEN TRIEU STEVE PLIMSOLL

Co-Founder Assistant CEO
CFTE National Center for AI

SPEAKERS

FATIMA ABUKAR AHMAD ALAJLAN DR. MOHAMMED MANSOUR ALSALEH

Principal Advisor, Manager of Fintech Permit ALHUSSEIN Head of Innovation Hub
Responsible AI and Data Capital Market Authority Chief Executive Officer Saudi Central Bank (SAMA)
Financial Conduct Authority MOZN

DOUGLAS ARNER MASSIMO BUONOMO DR. SIVARAMAKRISHNAN SOPNENDU MOHANTY

Kerry Holdings Expert and Project evaluator GURUVAYUR Chief Fintech Officer
Professor in Law for top AI and deep-tech Center for Responsible AI Monetary Authority
The University of European projects (CeRAI) - Research Advisor, of Singapore
Hong Kong European Commission & Chief AI Scientist
Aaquarians.ai

DR. MOHAMAD SIBAI RYOSUKE USHIDA

Chief Technology Officer Financial Services
Srena Agency of Japan
Chief Fintech Officer

3
Contents
Governance of AI Technologies
in the Financial Services

2 11
Preface PART 2
AI Governance Across Regions:
Key Objectives and Approaches

3 16
Contributors PART 3
Regulatory Objectives and
Emerging Challenges

5 20 28
Executive PART 4 Conclusion
Summary Balancing Ethical Concerns
with Innovation in AI

7 24 29
PART 1 PART 5 Appendix
Are we getting close to Future of AI Governance – List of Researchers
future-proof AI regulations? Preparing for Advanced AI

4
Executive Summary

A decade ago, the financial services sector was only beginning to explore the potential of AI,
with expectations of gradual, incremental integration. Today, however, AI has rapidly evolved
into a transformative force, with its impact accelerating faster than anticipated, especially
within financial services. The “Governance of AI technologies in financial services” report
offers vital insights into AI’s evolving role, addressing regulatory challenges, ethical dilemmas,
and the pressing need for collaborative governance frameworks.

The report summarises key discussions among industry leaders, regulators, and AI experts,
highlighting the opportunities and complexities AI brings to financial services. It emphasises
the urgent need for governance models that can balance innovation with ethical responsibility.

While some experts are optimistic about the progress of AI governance, many argue that
existing frameworks fall short of managing AI’s rapid evolution. This divergence reflects
the challenge of regulating AI systems capable of learning, adapting, and autonomously
influencing critical financial decisions.

Key findings of the report include:

AI Regulations Are a High-Priority Topic

The report highlights the significant engagement from all key stakeholders – regulators, AI
developers, and financial institutions – around AI regulation. With AI systems becoming more
integrated into financial processes, the need for robust and flexible regulations has attracted
wide attention, making it a top priority across the industry.

Future-proof AI Regulations Are Elusive

While progress has been made in developing AI regulations, many frameworks remain reactive.
The report emphasises the need for adaptive governance models that can evolve alongside AI
technologies to ensure they remain relevant in the face of future advancements.

Diverging Regional Approaches

The report highlights significant differences in how regions such as the EU, UK, Saudi Arabia,
and Singapore approach AI regulation. These variations reflect different priorities, with some
regions focusing on innovation and flexibility, while others emphasise consumer protection and
stricter ethical standards.

Ethical AI Is Paramount

One of the most pressing concerns is ensuring that AI systems are transparent, accountable,
and free from bias. The roundtable stressed the importance of embedding ethical
considerations into AI frameworks, particularly in high-risk areas like credit scoring and lending.

The Role of Regulatory Sandboxes

AI-specific regulatory sandboxes have emerged as a critical tool for fostering innovation while
ensuring regulatory oversight. These controlled environments allow financial institutions and
regulators to collaborate on testing AI technologies, helping shape effective governance.

5
Executive Summary

The Skills Gap

A critical issue raised is the need for upskilling both industry professionals and regulators. As AI
systems become more complex, financial institutions and regulatory bodies must ensure they
have the expertise to manage and govern AI-driven solutions effectively.

Collaboration Is Key

The report underscores the importance of cross-sector collaboration, involving regulators,

industry professionals, academia, and policymakers to create cohesive, globally aligned AI
governance frameworks.

The report concludes that the governance of AI in financial services requires a multi-stakeholder
approach. Effective AI governance must be flexible, inclusive, and future-ready, ensuring
that the industry can harness AI’s potential while safeguarding against its risks. This report
offers a foundation for ongoing dialogue and future research, urging leaders to prioritise both
innovation and ethical governance as AI continues to reshape the financial landscape.

6
PART 1

Are we getting close to

future-proof AI regulations?

7
Are we getting close to
future-proof AI regulations?
The roundtable discussion on AI governance began with a critical question: Are we getting
close to future-proof AI regulations? This sparked a lively and urgent debate, as experts
tackled the challenge of managing a technology that evolves faster than most frameworks
can keep pace. Some expressed optimism, highlighting recent strides toward adaptable,
forward-looking policies. However, others warned that many current regulations remain
reactive, struggling to keep up with AI’s rapid advancements.

Yes No
40% 60%

Figure 1: Expert Opinions on Whether We Are Close to Developing Future-Proof AI

Yes, Progress Is Being Made

Some speakers were optimistic, highlighting that although AI regulation is in its early stages,
significant progress is being achieved through international initiatives. These frameworks are
gradually aligning with local contexts, laying the groundwork for comprehensive, future-proof
AI regulations. However, they noted this progress is still in its “baby step” phase, requiring
ongoing adaptation to keep up with technological advancements.

No, Current Frameworks Are Insufficient

Others expressed scepticism, particularly in regard to the European Union’s regulatory efforts.
While the EU AI Act is a positive step, speakers pointed out its lack of specificity in addressing
the nuances of AI applications. Some felt that current frameworks are too rigid and lack the
flexibility needed to govern the full complexity of AI.

The Four Pillars of Regulatory Focus

One speaker introduced a helpful framework to understand the regulatory landscape, breaking
it down into four key areas: macro issues, sector-specific challenges, abuse of AI, and cultural
and human capital considerations. These areas illustrate the diverse concerns AI regulation
must address, highlighting that no single regulatory framework can comprehensively manage
every aspect of AI governance.

8
Are we getting close to
future-proof AI regulations?

Macro Issues Sector-Specific

Challenges

The Four Pillars

of AI Governance

Cultural and
Abuse of AI Human Capital
Considerations

Figure 2: The four pillars of AI governance

Macro Issues

AI’s potential to reshape employment, infrastructure, and data management raises broad
societal concerns, including its impact on national strategies and the singularity. These issues
affect all sectors and the economy as a whole.

Sector-Specific Challenges

Industries like financial services and healthcare face unique AI-related concerns, such as
fairness, data privacy, and preventing abuses. AI systems must be explainable, responsible, and
compliant with regulations specific to each sector.

Abuse of AI

The misuse of AI for cybercrime, fraud, and illegal activities is a critical concern. Effective
regulation requires cybersecurity frameworks to address malicious AI use, as traditional
regulations may not suffice.

Cultural and Human Capital Considerations

Society’s adaptation to AI includes education and upskilling, ensuring individuals are prepared
to work alongside AI systems and adapt culturally to new technologies.

9
Are we getting close to
future-proof AI regulations?

The Future-Proofing Dilemma

One speaker argued against the idea of “future-proof” regulations, emphasising that AI’s
rapid innovation makes it impossible to create long-lasting rules. Instead, they advocated
for adaptable, responsive regulatory frameworks that manage risks without stifling
innovation. Overly rigid or prescriptive regulations could lead to instability, particularly if
they change unpredictably.

Existing Legal Frameworks: Do We Need New AI Laws?

The discussion questioned whether new AI-specific laws are necessary. One speaker
suggested that many AI issues could be addressed through existing regulations, similar
to how current laws handle horse-related matters (the “law of the horse” analogy).
However, they acknowledged that as AI evolves, new regulatory elements may be
needed, especially for emerging risks like generative AI.

In summary, speakers agreed on the need for adaptable, flexible regulations that support
AI innovation while managing risks. The challenge is balancing technological growth with
regulatory protection.

The Law of the Horse

The “Law of the Horse” is a metaphor used in legal studies, particularly in the context
of cyberlaw, that critiques the idea of creating a specialised legal discipline for rapidly
evolving technologies. This concept was famously discussed by Judge Frank H.
Easterbrook in a 1996 lecture at the University of Chicago, where he argued against the
idea of developing a distinct body of law for the internet, which he referred to as “cyberlaw.

Easterbrook suggested that there is no more a “law of the horse” than there is a “law of
cyberspace.” He argued that just as legal issues concerning horses should be addressed by
general principles of law (like property, contract, or tort law), so should issues arising from
the internet. He believed that creating a specialised area of law for every new technology
or domain is unnecessary and inefficient, as it would fragment legal education and
practice without adding substantive value.

In contrast, Lawrence Lessig, another prominent legal scholar, argued against

Easterbrook’s view. In his response titled “The Law of the Horse: What Cyberlaw Might
Teach,” published in the Harvard Law Review, Lessig contended that studying the internet
as a specific field could provide unique insights into the broader legal landscape and
reveal how laws adapt to technology. He believed that specialised study could illuminate
the unique problems and regulatory challenges posed by digital spaces, thus meriting a
distinct area of legal study and practice.

Source: Frank H. Easterbrook: Cyberspace and the Law of the Horse and Lawrence Lessig: The Law of the
Horse: What Cyberlaw Might Teach

10
PART 2

AI Governance Across
Regions: Key Objectives
and Approaches

11
AI Governance Across Regions:
Key Objectives and Approaches

Overview of Regulatory Approaches by Region

The roundtable highlighted various regional approaches to AI governance in financial services,

focusing on Saudi Arabia, the UK, the EU, Singapore, and Japan.

Saudi Arabia

A speaker highlighted that Saudi Arabia views AI as a national priority, with significant focus
placed on AI education and infrastructure development. The Kingdom’s approach spans from
education at all levels to building AI infrastructure and ensuring data sovereignty. Significant
investments in local infrastructure, including data centers from companies like Google,
Microsoft, and Oracle, address regulatory challenges related to cloud usage and cross-border
data flows. Key initiatives, such as the Saudi Data and AI Authority’s (SDAIA) National AI Ethics
Checklist and partnerships with UNESCO, support responsible AI use.

“We have a national AI

ethics checklist that
allows you to go through
and identify what are the
risks of what you’re doing
against a standardised
checklist.”

United Kingdom

The UK emphasises outcome-based regulation, applicable to emerging technologies like AI. As

noted in its AI Update, the FCA is enabling the safe and responsible use of AI that drives growth
and competitiveness within the sector. Its regulatory approach prioritises risk management
and consumer protection, with frameworks such as the Senior Managers and Certification
Regime (SM&CR) holding firms accountable for AI risks. The FCA’s Consumer Duty ensures AI
decision-making is fair and delivers good outcomes for consumers.

European Union

The EU’s AI Act emphasises consumer protection, transparency, and fairness, particularly in
financial services. It imposes strict requirements on AI applications, such as credit scoring,
ensuring they respect individuals’ rights. The regulation is detailed, especially for high-risk AI
systems in sectors like finance, where decisions can impact the impact assessment.

12
AI Governance Across Regions:
Key Objectives and Approaches
Singapore

Singapore has taken a leadership role in AI governance, particularly in the financial

sector. Through initiatives like Project Veritas, Singapore addresses the ethical challenges
of AI in banking by creating open-source frameworks to assist financial institutions
in embedding principles of fairness, ethics, accountability, and transparency (FEAT)
into their AI systems. A speaker shared, “We worked on an industry-led programme...
we opened up a GitHub account with open-source code where tech companies,
regulators, and banks worked together,” which demonstrates Singapore’s collaborative
approach to tackling AI’s ethical risks.

Fairness Ethics
• Accuracy • Beneficence
• Bias • Human-centered
• Model agnostic • User privacy
• Justice AI
Governance
Transparency
Accountability • Explainability
• Regulatory compliace • Justifiability
• Trustworthiness • Reproductibility

Figure 3: FEAT Principles, Monetary Authority of Singapore

Japan

Japan’s AI governance focuses on managing continuously learning AI models, particularly

in regulated sectors like finance. Model risk management is a key concern, as the evolving
nature of AI makes validation challenging. The financial sector emphasises verifying AI
models to ensure accuracy and fairness as they adapt to new data over time.

Commonalities and Differences in Regulatory Focus

Throughout the roundtable discussion, speakers identified several commonalities and

differences in how AI is governed within financial services across various regions. While there
was consensus on some key objectives, there were also significant points of divergence in
regulatory approaches.

13
AI Governance Across Regions:
Key Objectives and Approaches

Common Objectives

Consumer Protection

One of the most universally shared goals across regions was consumer protection, especially
in areas where AI is used to make decisions in sectors like lending, hiring, and credit scoring.
A common theme among the speakers was ensuring fairness and transparency in AI-driven
decision-making. “AI must assess the impact on fundamental rights,” emphasises the need
to safeguard consumers from biased or opaque algorithms. This focus is especially strong in
the EU and UK, where strict regulations and outcomes-based frameworks aim to mitigate
risks to consumers.

Risk Management

Speakers agreed that these AI models, which evolve over time, pose unique challenges for risk
management. The complexity lies in the fact that these models require constant revalidation
as they learn from new data inputs. This continuous evolution introduces additional layers of
oversight, making it difficult for regulators and financial institutions to ensure that the models
remain compliant and accurate.

Managing continuously learning AI models is complex... If

models evolve, they need constant revalidation, which adds
significant complexity to the risk management process.

Data Security and Sovereignty

Data security, particularly data sovereignty, is a key focus with regional variations. In Saudi
Arabia, data sovereignty is crucial due to concerns about cloud usage and localising data, with
efforts to build local AI infrastructure. Meanwhile, the EU emphasises data protection through
regulations like GDPR and the upcoming AI Act, ensuring careful handling of data in AI systems.

Key Differences

Regulatory Flexibility

Regions like the UK and Singapore favor flexible, principles-based approaches that adapt to
evolving AI technology, promoting innovation with oversight. In contrast, the EU and Japan
have stricter, more prescriptive regulations, especially for high-risk AI systems, offering greater
protection in sectors like finance where AI’s impact is significant.

14
AI Governance Across Regions:
Key Objectives and Approaches
Sectoral Focus

Debate arose over the focus of AI regulation. Japan emphasises AI model risk management
specifically within the financial sector, ensuring accuracy and fairness. In contrast, Saudi
Arabia adopts a broader approach, focusing on cross-sectoral AI infrastructure development.
Opinions differed on whether a focused or broad approach would deliver better long-term
results for AI governance.

Sector-Specific AI Regulations

Financial Services

AI’s growing role in areas like fraud detection, credit scoring, and algorithmic trading brings
significant regulatory challenges. Several speakers highlighted concerns about model
risk management, particularly with continuously learning AI models. The rapid pace of AI
development complicates the validation and verification of these models, especially as they
evolve beyond traditional rule-based systems, as seen in credit scoring.

Another speaker emphasised that, much like the statistical modeling techniques financial
services have long relied on before AI, AI-driven creditworthiness assessments require strict
oversight to ensure fairness and transparency. The speaker highlighted the importance of
preventing discrimination or biased outcomes, particularly in lending practices.

“You can’t leave AI

deciding whether you
are creditworthy or not”

A speaker highlighted that AI is moving towards

microservices and service-oriented models,
emphasising the need for financial services to
quickly adapt to these technological changes.

Data Privacy

The issue of data privacy sparked considerable debate among the speakers, with regional
differences coming to the forefront. In the EU, laws like GDPR heavily influence how AI
systems operate, enforcing strict data protection, especially across borders. While many
praised GDPR’s robust protections, some voiced concerns that its stringent rules could
hinder innovation by limiting data usage.

15
PART 3

Regulatory Objectives and

Emerging Challenges

16
Regulatory Objectives and
Emerging Challenges

Core Regulatory Objectives

During the roundtable discussion, speakers emphasised the core regulatory objectives
driving AI governance in financial services. While these objectives are well-established,
many speakers agreed that they need to be adapted to the unique complexities introduced
by AI-driven systems.

Know Your Customer (KYC)

AI is becoming crucial for verifying customer identities in financial services, enhancing

fraud protection and streamlining onboarding. While AI efficiently processes large data
volumes, speakers stressed the need for explainability and transparency, especially when
models flag or reject customers, ensuring security and accuracy without sacrificing
customer experience.

Anti-Money Laundering (AML)

AI plays a key role in detecting suspicious activities by identifying patterns in vast datasets.
While AI is powerful, speakers acknowledged its limitations in addressing emerging threats.
Ensuring AI systems adapt to new forms of financial crime while maintaining accuracy
remains a challenge.

“We know that there are some objectives which, especially

in financial services, will be the same – consumer protection,
financial stability, etc.”

Consumer Protection

Fairness, transparency, and non-discrimination

in AI-driven financial services were central
concerns. However, one voiced concerns that
AI systems, if not carefully designed, could
unintentionally perpetuate or even amplify
existing biases, particularly in sensitive areas
like lending and credit scoring. The Financial
Conduct Authority (FCA) emphasised that
AI systems must align with fairness and
transparency principles. The UK’s Consumer
Duty policy further reinforces the need for
accountability and ethical AI governance to
prevent discrimination and ensure equitable
access to financial services.

17
Regulatory Objectives and
Emerging Challenges

Market Integrity

AI in algorithmic trading presents both opportunities and risks. While it can make markets
more efficient, it also increases the potential for manipulation. One speaker emphasised
the need for robust regulatory frameworks to monitor AI systems and prevent
market destabilisation.

Financial Stability

The evolving nature of AI in financial decision-making poses risks to financial stability.

Self-learning AI systems can behave unpredictably, complicating risk management.
Speakers stressed the need for rigorous validation and oversight mechanisms to ensure AI
models do not introduce unforeseen risks, especially in highly regulated sectors like banking.

“So from a financial stability perspective, this concentration

risk and dependency on particular and very few providers
for AI services is actually something concerning.”

Emerging Challenges with AI Governance

The roundtable discussion highlighted several challenges in AI governance, reflecting the

complexities of AI’s rapid evolution in financial services.

Bias and Fairness

Speakers emphasised the risk of AI amplifying societal biases, especially in areas like lending,
hiring, and insurance. There was consensus on the need for strict regulatory safeguards to
prevent biased algorithms from perpetuating inequalities. One speaker stressed that AI
systems must have proper checks to avoid discriminatory practices.

Explainability

The increasing complexity of AI systems makes explainability a key challenge, especially in

finance where transparency is essential. Speakers agreed that AI must not be treated as a
“black box” and should be designed with mechanisms to explain decisions to regulators
and consumers, ensuring clarity and accountability in decisions like credit approvals or
fraud detection.

Continuous Learning Models

The oversight of continuously learning AI models, which adapt over time, was highlighted as
a significant challenge. Unlike static models, these systems evolve, making them harder to
validate and audit. A speaker noted the difficulty for regulators in ensuring these AI models
remain compliant with standards while fostering innovation.

18
Regulatory Objectives and
Emerging Challenges

Cross-Border Data Flows

The issue of cross-border data flows and data sovereignty was a recurring theme in the
discussion. AI systems often rely on large datasets, which can be sourced from multiple
countries, each with its own privacy and data protection laws. Differences in regulations, such
as the EU’s GDPR and Saudi Arabia’s data sovereignty laws, create barriers to the development
of unified AI frameworks. Speakers acknowledged that addressing these disparities is critical
for fostering international cooperation and enabling the safe and secure operation of AI
systems across borders.

Adapting Regulations to Rapid Technological Advancements

Future-Proofing AI Regulations

One of the central points of agreement was that AI regulations must be flexible and
adaptable. The speakers acknowledged that the AI of today will not be the AI of tomorrow,
with advancements like generative AI and quantum computing on the horizon. While some
advocated for a reactive approach, adjusting regulations alongside advancements, others
stressed that regulation should not hinder innovation. The key takeaway was the importance
of creating adaptable guidelines that balance innovation with the risks posed by emerging
technologies like quantum computing, ensuring regulations remain relevant as AI evolves.

“The AI that we know today is not going to be the AI that

we know tomorrow... I’m talking about the movement or
transition from AI based on GPUs and special purpose CPUs
to quantum computing and other emerging technologies.”

AI-Specific Regulatory Sandbox

The discussion highlighted the importance of AI-specific regulatory sandboxes, such as

those in Saudi Arabia, allowing AI technologies to be developed and tested in a controlled
environment. This approach helps balance innovation and regulation, enabling both
companies and regulators to learn and assess risks. One speaker highlighted that regulatory
sandboxes provide valuable insights into AI performance in real-world scenarios, aiding in
shaping future regulations. However, challenges like limited data availability in Saudi Arabia’s
banking ecosystem were mentioned, with a suggestion to create shared synthetic data assets
for training AI models.

19
PART 4

Balancing Ethical Concerns

with Innovation in AI

20
Balancing Ethical Concerns
with Innovation in AI

Ethical Challenges in AI Development

During the roundtable, ethical concerns in AI development, particularly within financial

services, were central to the discussion. These concerns highlight the complex trade-offs
between innovation and the need to maintain fairness, transparency, and accountability in
AI-driven systems.

Bias and Transparency

Fairness & Explainability

Accountability

Figure 4: Ethical challenges in AI development within financial services

Bias and Fairness

Speakers agreed that AI systems risk reinforcing societal biases, particularly in financial areas
like lending and hiring. While AI can improve efficiency, it may perpetuate discrimination
if not properly regulated. Safeguards are essential to ensure fairness and prevent biased
algorithms from harming marginalised individuals.

“If you think about how these models reinforce bias in

credit scoring or lending decisions, the issue isn’t just about
biased data but also how these systems create an unfair
disadvantage for individuals who are already marginalised.”

21
Balancing Ethical Concerns
with Innovation in AI

Transparency and Explainability

The lack of transparency in AI decision-making was another major concern. In areas

like lending and credit scoring, AI often operates as a “black box,” making it difficult
to understand how decisions are made. Speakers emphasised the need for regulatory
frameworks that ensure AI systems provide clear, understandable explanations, particularly
for adverse decisions.

Accountability

Assigning responsibility for harmful AI decisions was also debated, as AI’s complexity
complicates accountability. Financial institutions must establish clear accountability lines
to address errors or biases. Regulating accountability for evolving AI models is a significant
challenge that must be addressed to protect consumers and maintain ethical standards.

Encouraging Responsible Innovation

Despite ethical challenges, the roundtable emphasised the importance of fostering

responsible innovation in AI for financial services, provided ethical safeguards are in place.

Regulatory Sandboxes

Building on the earlier discussion of AI-specific regulatory sandboxes (page 19), speakers
further emphasised their role as key tools for enabling AI innovation while maintaining
oversight. These controlled environments allow AI technologies to be tested and developed
under regulatory supervision, balancing innovation with consumer protection and financial
stability. Sandboxes provide opportunities for regulators and developers to learn and adapt,
ensuring regulations stay relevant. However, there was caution that sandboxes should not be
overly restrictive, as this could hinder innovation.

Public-Private Partnerships

The importance of public-private partnerships was another key theme. Collaboration

between regulators, academia, and AI developers was seen as essential for aligning
innovation with ethical standards. These partnerships allow for knowledge exchange,
ensuring regulators stay informed on technological advancements while developers gain
clarity on compliance, fostering the development of adaptable, ethical frameworks.

22
Balancing Ethical Concerns
with Innovation in AI

New Challenges in AI Ethics

As AI technologies continue to evolve, new ethical challenges emerge that require ongoing
attention and adaptation by regulators and industry stakeholders.

Model Risk and Bias

The complexity of monitoring AI models for bias, especially as they continuously learn, was
a key concern. While AI can streamline decision-making, it risks reinforcing societal biases,
particularly in areas like lending and credit scoring. Speakers emphasised the need for
ongoing monitoring and retraining to ensure fairness, as failure to address biases could lead to
discriminatory practices in financial services.

Ethical Enforcement

Ensuring ethical compliance was another challenge. While regulations exist, speakers stressed
the importance of enforceable guidelines. Developing effective mechanisms for auditing AI
systems and holding parties accountable for ethical breaches is crucial, especially given the
autonomous nature of AI. Robust frameworks for audits and oversight are needed to ensure
compliance with ethical standards.

Building an AI-Ready Workforce

The discussion also underscored the importance of building a workforce capable of navigating
the ethical and regulatory challenges posed by AI technologies.

Upskilling Regulators

The discussion stressed the need for regulators to gain AI-specific expertise, particularly in areas
like machine learning and continuous learning models. Without this knowledge, regulators
may struggle to oversee rapidly evolving AI systems, risking unchecked use in financial sectors.

“Very quickly. It’s a talent development so it’s a sure

challenge for regulators like FSA...we are working together
with academia or some industry expert as a kind of advisor.”

One of the speakers highlighted the challenge of ensuring that regulators not only
understand the technical aspects of AI but are also equipped to apply this knowledge in
real-world oversight scenarios. Several speakers emphasised the importance of training and
development programmes that allow regulators to keep up with the evolving nature of AI
technologies. It was broadly agreed that ongoing education is critical for effective governance,
with a focus on bridging the knowledge gap between technologists and policymakers.

Collaborations with Industry and Academia

Collaboration between industry, academia, and regulators is crucial for building an AI-ready
workforce. Initiatives like Saudi Arabia’s AI education programmes serve as models for training
professionals to develop and regulate AI systems responsibly, aligning with ethical standards
and regulatory requirements.

23
PART 5

Future of AI Governance –
Preparing for Advanced AI

24
Future of AI Governance –
Preparing for Advanced AI

Expectations for AI Advancements

The roundtable continued with Ray Kurzweil’s concept of the Law of Accelerating Returns,
as presented in his book The Singularity Is Near: When Humans Transcend Biology. Kurzweil
predicts that the 21st century will see 1,000 times the progress achieved in the 20th century
due to the exponential growth of technology.

Uncertainty Around AGI

The discussion began with differing views on whether superintelligence could emerge within
the speakers’ lifetimes. While many felt artificial general intelligence (AGI) was becoming
more plausible, driven by advances in AI like large language models (LLMs), uncertainty
remained about future developments. One speaker suggested that traditional regulation
might be inadequate for AGI, proposing a constitution for AI as a legal framework to guide
governance. However, skepticism was voiced about current AI capabilities, with concerns
about the influence of large corporations over AI’s development, making effective governance
challenging despite regulatory efforts.

Artificial Super
Intelligence

Artificial General
Intelligence AI represents
intelligence that
Current surpasses human
Breakthroughs AI is capable capabilities
Artificial Narrow of performing
Intelligence multiple tasks at a
WE ARE human level

HERE
AI is designed to
handle simple,
single-task activities
with high efficiency

Source: IDB Invest

Ethical Considerations and Long-Term AI Risks

As AI becomes more advanced and integrated into critical sectors such as finance, its ethical
implications become increasingly pressing. The roundtable also highlighted several long-term
ethical risks that will need to be addressed as AI technology continues to evolve.

25
Future of AI Governance –
Preparing for Advanced AI
Autonomy and Accountability

The discussion emphasised the growing autonomy of AI systems and the need for human
oversight, especially in sectors like finance and healthcare. As AI takes on more decision-
making, speakers highlighted the increasing risk of ethical decisions being left to machines.
The need for clear accountability frameworks was stressed, ensuring humans remain
responsible for AI-driven decisions. Without such frameworks, assigning liability for errors or
unethical outcomes from autonomous AI will be difficult.

“The number of
machine-to-machine
financial transactions
will increase significantly
if we have something
like AGI and AI.”

Global Ethical Standards

Speakers emphasised the critical need for international cooperation to establish consistent
ethical standards for AI governance. While much of the current regulatory activity occurs at the
national government and regulator level, there is significant global momentum to harmonise
these efforts. Initiatives like the Bletchley Declaration, the World Bank’s AI governance
programme, and standards from organisation such as the National Institute of Standards
and Technology (NIST), with its AI 100-5, A Plan for Global Engagement on AI Standards, and
the International Organisation for Standardisation (ISO), through its AI management system
standard ISO/IEC 42001, are shaping the global dialogue on AI ethics. These global guidelines
and initiatives are essential for promoting fairness, transparency, and accountability, particularly
in industries like financial services, where AI’s impact crosses borders.

While global standards often align with national efforts, there can be discrepancies where
international guidelines do not immediately match specific national views or priorities. However,
initiatives such as the G7 Hiroshima Process and the Transatlantic Trade and Technology Council
underscore the importance of cross-border collaboration to develop a unified framework. These
collective efforts are crucial for preventing regulatory arbitrage, ensuring responsible innovation,
and fostering a globally consistent approach to the ethical use of AI.

26
Future of AI Governance –
Preparing for Advanced AI

Practical Challenges of Future AI Governance

Scaling AI Infrastructure

As AI systems advance, scaling infrastructure, particularly in sectors like finance and healthcare,
is crucial. Speakers agreed that the growing complexity of AI models requires
significant investments in computational resources, data storage, and network capabilities.
Smaller markets, like Saudi Arabia, face unique challenges in keeping pace with global
developments. Regulators will need the capacity to oversee these complex systems, with some
speakers raising concerns about local capabilities compared to larger markets.

Concentration of Power

The growing reliance on a few major tech companies for AI services, such as cloud computing,
was another key concern. This concentration of power could pose systemic risks, especially in
smaller markets where dependence on external providers increases vulnerability. One speaker
noted, “Big tech companies offer AI capabilities that SMEs can’t access,” highlighting the
financial stability risks if these providers face disruptions. To mitigate this, some emphasised
the need to build local AI capabilities and reduce reliance on multinational firms.

“Big tech companies offer AI capabilities that SMEs can’t

access... There’s a dependency on very few providers for AI
services, which poses a financial stability risk.”

AI Opinion Shaping and Monopolies

A significant concern was the potential for AI, particularly generative AI, to shape opinions in
sectors like finance and insurance, driven by monopolistic organisations. This manipulation of
decision-making processes could become a major challenge as AI’s influence grows.

Web 3.0, Data Nationalism, and AI Integration

Speakers discussed the potential for Web 3.0 technologies to decentralise financial systems,
adding complexity and risk. However, the rise of data nationalism – countries enforcing stricter
data boundaries – could counterbalance these changes. Privacy-preserving technologies could
allow secure cross-border data sharing, though this is still developing. Concerns were also
raised about the significant energy consumption of AI technologies, with data centers expected
to account for two-thirds of energy resources by 2030 if unchecked.

27
Conclusion

The roundtable discussions highlighted the undeniable need for transformation in

workforce development as AI, automation, and digital technologies reshape industries at an
unprecedented rate. To remain competitive and innovative, organisations must act now to
bridge the skills gap and foster a workforce that is not only proficient in technical skills but
also adaptable and emotionally intelligent.

Key themes that emerged throughout the discussions underline the importance of cross-sector
collaboration, continuous learning, and diversity in building the workforce of the future. The rise
of hybrid skills, where technical proficiency is complemented by soft skills like critical thinking,
adaptability, and leadership, is paramount to ensuring employees are equipped to navigate
complex challenges in an AI-driven world.

Moreover, the role of leadership in steering this transformation cannot be overstated.

Adaptive leadership, rooted in transparency, empathy, and support, is essential for creating an
environment that encourages experimentation, learning from failure, and resilience in the face
of rapid technological change. Leaders must lead by example, fostering a culture of innovation
and lifelong learning.

In conclusion, building the workforce of tomorrow requires a proactive and holistic approach.
Organisations that invest in reskilling and upskilling, embrace diversity, and cultivate strong
leadership will not only thrive in the evolving digital landscape but will also ensure that their
workforce remains agile, resilient, and prepared for the future. This report lays the foundation
for ongoing dialogue and action, urging leaders and stakeholders to prioritise workforce
transformation as a strategic imperative.

28
Appendix

29
Researchers

Content and Research Marketing and Design

Huy Nguyen Trieu Maryam Ahmad Noor
Co-founder Fintech Analyst

Tram Anh Nguyen Huong Giang Dinh

Co-founder Fintech Research Analyst

Maryam Ahmad Noor

Fintech Analyst

Huong Giang Dinh

Fintech Research Analyst

30
Centre for Finance,
Technology and
Entrepreneurship
Founded in 2017 in London, CFTE is a global platform
for education in Fintech and the future of Financial

More than 100,000 professionals from 100+ countries In total, more than 200 CFTE experts provide a
have participated in CFTE programmes to accelerate global view of what’s really happening in this new
their careers in Fintech and new finance. In addition world of finance.
to London, CFTE is present in Singapore (accredited
by Institute of Banking and Finance), Abu Dhabi “In a tech world, we bet on people” is CFTE’s
(Abu Dhabi Global Market Academy), Hong Kong motto. Our global community is the core of
(Cyberport), Malaysia (Asian Banking School), CFTE. Thanks to an innovative and open mindset,
Luxembourg (Luxembourg Academy of Digital CFTE alumni progress in their careers and help
Finance with LHOFT) and Budapest (Budapest others do the same, with notable alumni leading
Institute of Banking). transformation in their organisations. They also
attend events and share advice, tips and job
CFTE’s objective is to equip professionals and opportunities. CFTE alumni have also made an
students with the skills to thrive in the new impact through the world’s largest Global Fintech
world of finance. This includes online courses and Internship by mentoring over 1,000 students from
specialisations, leadership training and hands- all over the world.
on extrapreneurship experiences in topics such
as Fintech, Open Banking, Digital Payments and CFTE believes that the new world of finance will
Artificial Intelligence. be inclusive, diverse, innovative and will have a
positive impact on society and people. This starts
CFTE courses are designed with the principle of with people having the right knowledge and
For the industry, By the Industry. Our courses are mindset so that no one is left behind. Whether
taught by senior leaders from fast-growing Fintech you want to learn, contribute or more generally be
companies such as Revolut, Plaid, and Starling Bank, part of the new world of Financial Services, we are
innovative financial institutions such as Citi, DBS and looking forward to welcoming you.
Ping An, tech companies such as Google, IBM and
Uber and regulators from MAS, ECB and MNB.

Contact Website
Research team Courses
[email protected] courses.cfte.education

Press Articles
[email protected] blog.cfte.education

31
About
Fintech Saudi
Fintech Saudi is an initiative launched by the Saudi Central Bank
(SAMA) in collaboration with the Capital Markets Authority (CMA) under the Financial Sector
Development Programme to support the development of the Fintech Industry in Saudi Arabia.
Fintech Saudi’s ambition is to transform Saudi Arabia into an innovative fintech hub with a
thriving and responsible fintech Ecosystem.

Fintech Saudi seeks to achieve this by supporting the development of the infrastructure
required for the growth of the fintech industry, building capabilities and talent required by
fintech companies and supporting fintech entrepreneurs at every stage of their

We Develop We Build Capabilities We Support

Supporting the Building the skills and Supporting fintech

development of the knowledge required entrepreneurs at
infrastructure required for the growth of the every stage of their
by the fintech industry fintech activity development

Signature
Activities

Initiatives

Content

32
More from CFTE

Fintech Job Blockchain job SHIME Skills

Report Report Framework

Get in touch with us

Business Development Partnerships
Faheem Yunius Melike Evtut
[email protected] [email protected]
Head of Growth, CFTE Europe

Piotr Kurzepa
[email protected]
[email protected]
Head of Middle East Office

33