AI’s Impacts on Regulation

Mr. Tanut Puangnuam
Ms. Peechanika Phoomrittikul
Mr. Ratchapol Pongprasert

1. Introduction

On 19^th February 2026, the Local Reach Programme hosted a session focused on the transformative impact of Artificial Intelligence (AI) on the regulatory landscape. This event, a collaboration between the Macroeconomic Policy Division of the Fiscal Policy Office and the Finance and Fiscal Journal, featured Mr. Kenn Rodrigues, CEO and Co-founder of DHI-AI (Decision through Holistic Intelligence).

As the leader of an Australian-based AI firm specializing in advanced solutions for data extraction and market oversight, Mr. Rodrigues shared practical insights on how organizations are moving away from manual, resource-constrained processes toward autonomous systems that can keep up with today’s complex markets.

Mr. Kenn Rodrigues
CEO and Co-founder of DHI-AI

2. The Emergence of AI

The regulatory environment is currently navigating the rapid rise of several sophisticated technologies:

Generative AI & LLMs: Large Language Models are becoming foundational for processing and structuring vast amounts of unstructured data.
Autonomous Agentic Systems: These systems are utilized for autonomous evidence collection and investigation workflows, significantly reducing manual labour and increasing oversight efficiency.
Vector Databases & Semantic Search: These tools enable contextual, high-speed searches across massive, unstructured datasets to uncover hidden insights.
Real-time Anomaly Detection: By utilizing Bayesian networks and agentic systems, regulators can move beyond rigid, rules-based monitoring to identify “unknown-unknown” risks and behavioral deviations as they occur.
Multimodal AI: These models integrate and analyse text, voice, image, and video to detect complex fraud, such as deepfakes, manipulation, and misrepresentation.

While these tools offer immense potential for “good actors,” such as regulators and stock exchanges, to enhance market integrity, they are equally available to “bad actors” seeking to circumvent the rules. Consequently, it is critical that regulators continuously adapt their frameworks to maintain a technical advantage in an increasingly automated market.

3. Regulatory Challenges and Practical Applications of AI in Regulatory Intelligence

Market regulators globally—including those in the UK, Australia, New Zealand, Canada, and Malaysia—are currently evaluating alternative capabilities to address several challenges:

Resource Constraints: Monetary and personnel limits hinder the ability of regulators to effectively supervise and enforce compliance across expanding markets.
Increasing Market Complexity: The evolution of equity markets and intricate company structures has made traditional oversight significantly more difficult.
Data Volume Explosion: The sheer amount of trade data, regulatory requirements, and mandatory reports has reached a point where manual oversight is no longer feasible.
AI Capability Gap: There is a growing disparity between the advanced AI tools used by the industry and the current technical capabilities of regulatory bodies.
ESG Reporting & Monitoring Pressures: There is a heightened demand for regulators to ensure the integrity and validation of environmental, social, and governance (ESG) data.

To bridge the existing capability gap, regulators and market participants are deploying specific AI solutions to automate certain aspects of the compliance lifecycle and to enhance oversight:

Autonomous Evidence Collection: Intelligent agents are utilized to independently gather and structure case evidence, replacing slow, manual data collection processes.
False Positive Reduction: Through “agentic prioritisation” and intelligent alert triage, AI significantly reduces the noise of the traditional rule-based systems, allowing investigators to focus on high-probability threats.
Real-time Anomaly Detection: By utilizing Bayesian networks, regulators can move beyond rigid, rules-based monitoring to identify “unknown/unknown” risks and behavioral deviations as they occur.
Structured Data Extraction (SDE): Automated ingestion of financial and non-financial reporting ensures real-time market transparency and highly accurate data analysis especially for unstructured datasets.
Multimodal Fraud Detection: Advanced voice and image AI tools are specifically deployed to detect sophisticated manipulation, misrepresentation, and fraud, such as deepfakes.

4. Case Study: The Australian Securities and Investments Commission (ASIC)

The stock exchange previously faced significant hurdles with manual data extraction and an inability to continuously monitor market events, which severely impacted the efficiency and accuracy of their surveillance. To address these challenges, DHI-AI deployed a four-pillar approach to automate and enhance their oversight capabilities:

Image 1: DHI-AI Approach

A critical component of the AI Structured Data Extraction workflow is the “human-in-the-loop” protocol, which ensures that automated efficiency never compromises regulatory precision. While the system achieves an average 95% of accuracy on its first pass, the DHI platform is designed with a dedicated interface for manual validation.

As shown in the platform interface below, analysts can review AI-extracted data points—such as financial figures from a Director’s Report—directly alongside the original source document.

Image 2: Data Extraction and Structuring

Measurable Results and Benefits

The impact of moving from manual processes to an AI-led system was profound:

Efficiency: Achieved a 90% increase in the efficiency of data processing and analysis.
Speed: Reduced the time for report generation from 10 days to just 1 day.
Information Retrieval: Accelerated data searches from half a day to 5 minutes through intelligent agents.
Accuracy: Maintained a 95% of average extraction accuracy on the first pass prior to human-in-the-loop validation

4. Strategic Roadmap: Strengthening Thailand’s Regulatory Future

The session concluded with a robust Q&A, offering strategic guidance for government agencies, particularly in Thailand:

The Imperative for Government Adoption: Regulators must adopt AI for speed—to keep pace with rapidly moving modern markets and to effectively counteract malicious actors already weaponizing AI.
Cross-Border Implementation Challenges: Deploying AI in diverse regions introduces language barriers, requiring models to accurately interpret non-English text. Additionally, data sovereignty and security are paramount concerns.
Requirements for Thailand: To build local AI capacity, Thailand needs robust local programming talent and deep domain expertise. However, the most critical prerequisite is access to clean, machine-ready data.
Government’s Role in Ecosystem Building: Mr. Rodrigues highlighted Victoria, Australia, as a model. Programs like LaunchVic and Global Vic actively facilitate collaborations, connect startups with Venture Capital (VC) funding, and support the export of domestic technological innovation.
Detecting ESG Greenwashing: AI tackles ESG anomalies by validating a company’s internal sustainability claims against external data sources (such as news and third-party reports) to detect misrepresentations.

5. Conclusion

Looking ahead, Mr. Rodrigues predicts that AI will become increasingly autonomous and deeply embedded into daily regulatory workflows. However, this transition is no longer just a choice—it is a strategic necessity to stay ahead of the increasing governance requirements of the market. In addition, “bad actors” are increasingly leveraging AI to bypass market rules.

To ensure long-term market stability, the session highlighted three final pillars for success:

The Human Foundation: AI is designed to augment, not replace, human judgment.
A human-in-the-loop approach remains essential to validate AI outputs and ensure ethical accountability.
Data Readiness: The most critical prerequisite for Thailand’s AI journey is the availability of clean, machine-ready data. Without high-quality data, even the most advanced AI cannot function effectively.
Collaborative Ecosystems: Success requires more than just technology; it requires a community. Thailand can look to models like Australia’s LaunchVic to foster partnerships between the public sector, startups, and industry to accelerate innovation.

Ultimately, by moving from manual processes to intelligent oversight, agencies can achieve the vision of creative policy for modern organizations, ensuring a sustainable and resilient economy in the digital age.

We extend our sincere gratitude to Mr. Rodrigues for graciously accepting this opportunity and sharing his expertise with us. His insights will provide a perspective on how AI can be used to enhance public sector oversight and market stability. We would also like to express our appreciation to Dr. Norabajra Asava-vallobh, Director of Economic Data Innovation and Research Division and Mr. Mayoon Boonyarat, Director of Social Protection Strategy and Development Division at the Fiscal Policy Office for organizing this Local Reach Programme.