18/06/2026

BIZ & FINANCE THURSDAY | JUNE 18, 2026

20

organisations still manage fraud reactively: an issue surfaces, an investigation is launched, and management scrambles to contain the fallout, often with a keen eye on the quantum of loss and the cost of the investigation. Yet the real cost extends far beyond the financial loss – it includes operational disruption, reputational damage, erosion of stakeholder confidence, and the inevitable leadership question: “How did we not see this earlier?” In today’s fast-paced, data-driven environment, that question is becoming harder to justify. In most cases, warning signs already existed— but were not detected in time. A shift is underway. The Association of Certified Fraud Examiners (ACFE) consistently identifies tips as the leading method of fraud detection, while over half of frauds stem from control gaps or overrides—indicating that the issue frequently lies in control execution, not simply “bad people”. As a result, organisations are moving ACROSS Malaysia’s mid- to large scale manufacturing facilities, the imprint of Industry 4.0 is increasingly visible on the factory floor. From sensors on production lines to AI cameras at entry points to dashboards glowing in control rooms. On paper, the transformation is well under way and aligned with national initiatives such as the New Industrial Master Plan 2030 (NIMP 2030), the Malaysia Digital Economy Blueprint, and DoSH ((Department of Occupational Safety and Health)’s Vision Zero. However, discussions with plant managers point to a more complex reality. Data related to near-miss incidents often remains fragmented— camera footage is stored in one system, maintenance records in another, while safety reports continue to be compiled manually at the end of the week. The visibility exists, but it isn’t connected. At nearly 4% of global GDP, the cost of workplace incidents, as estimated by the International Labour Organisation (ILO), points to a structural issue, one that raises the question: Is integrated safety the missing piece in Malaysia’s Industry 4.0 ambitions? A digitised factory floor is not the same as an intelligent one Malaysia’s workplace safety record reflects a system under pressure. According to a report by Bernama, Malaysia recorded 4,409 workplace accidents between January and May 2025, with the manufacturing sector being the highest contributor. Mohd Hatta Zakaria, director-general of DoSH, has highlighted the disproportionately high incidence of workplace accidents originating from industrial corridors in Johor, Perak, and Penang. This indicates a pattern that points not to a lack of safety awareness, but to a structural gap in how safety is being monitored and managed. The irony is that many of these facilities have already made significant digital investments. The problem here is not a shortage of data, but a shortage of connected

Integrated AI safety key to success of Industry 4.0

beyond reactive responses towards continuous assurance, integrating proactive Fraud Risk Monitoring (FRM) with trusted Whistleblowing (WB) frameworks to strengthen early detection, accountability, and organisational resilience. For decades, organisations have relied on internal controls and audit cycles to manage fraud risk. While these mechanisms remain important, they have inherent limitations. Periodic audits detect issues only after they have occurred, sampling methodologies leave significant amounts of data unexamined, manual processes struggle to keep pace with growing transaction volumes, and static controls can be bypassed or overridden. In a business environment where thousands or even millions of transactions occur daily, these approaches are no longer sufficient. Fraud is no longer occasional or obvious; it is faster, more connected, and often disguised within “normal” activity. This is where FRM changes the equation: it delivers continuous visibility. By leveraging data analytics and control signals, FRM provides ongoing, real-time visibility across key risk areas workplace safety data. In one example cited by viAct, a large construction operator continued to face recurring safety challenges. Near-miss incidents involving heavy machinery, risks around open edges and confined spaces, and PPE non compliance were not consistently identified through manual audit processes. At the same time, digital tools in use operated independently: a computer vision system monitored PPE compliance, while incident reporting was managed through a separate platform. Following the deployment of an integrated AI safety system, the same site reported a 10x improvement in safety scores, alongside stronger alignment with regulatory compliance requirements. Each of these systems performed its intended function, but operated in isolation, without a unified view of risk. As a result, safety and productivity continued to be managed as parallel priorities handled by different teams, across disconnected platforms, even within facilities that had already considered themselves digitally mature. Why Vision Zero in Malaysia actually needs Integrated AI Safety Systems DoSH’s Vision Zero initiative is Malaysia’s national commitment to eliminate workplace fatalities and permanent disabilities, and is frequently cited as a policy target. But operationally, it sets a far higher bar than most safety systems on the ground are built to meet. What Vision Zero demands is a live operational awareness of risk , which an integrated AI safety

tools analyse transaction patterns, pricing anomalies, and payment flows 0 A whistleblower flags unusual payment practices monitoring detects duplicate payments, approval overrides, or timing irregularities 0 Concerns about conflicts of interest arise data analytics identify hidden relationships or recurring transactions This integration enables organisations to validate concerns quickly using data; identify broader patterns beyond the initial allegation; reduce investigation time and cost, and strengthen evidentiary support for decision-making. FRM without human input may generate false positives. WB without monitoring may depend too heavily on individuals speaking up. Together, these mechanisms create a more resilient and responsive control environment, enabling organisations to intervene earlier—often before issues escalate into material losses, regulatory investigations, or public crises. This article is contributed by BDO Malaysia executive director (advisory) Shirley Tey Sheh Lee. afternoon is no longer a coincidence waiting to be discovered in a monthly safety audit. It is a live operational signal, surfaced in real time, addressable before the shift ends. Gary Ng, CEO of viAct, puts the premise plainly. “The manufacturers who will define Malaysia’s next industrial chapter are not asking whether to invest in safety AI or productivity AI. They are asking why those should ever have been two separate decisions. When you bring safety and operational data onto a single intelligence layer, you stop managing incidents and start preventing them.” What this AI integration looks like on a real factory floor Consider a scenario familiar to any plant manager in Malaysia’s food and beverage sector. A worker on a high speed assembly line begins skipping a hand hygiene checkpoint to maintain pace during a peak production window. In a fragmented system, this surfaces, if it surfaces at all, as a compliance violation in the safety log at the end of the shift. In viAct’s integrated AI environment, real-time alerts do more than flag violations. They reveal recurring breaches on specific production lines, during specific shifts and under specific supervisors. That pattern is the intelligence. It tells the operations team that the process design is creating pressure that the safety procedure was not built to absorb, and it gives them the data to redesign the workflow rather than simply discipline the behaviour. This article is contributed by viAct CEO and co-founder Gary Ng ( pix) .

platform like viAct brings in 2026. These systems do not simply detect violations; they identify patterns across thousands of micro-events like near misses, repeated unsafe behaviours, and shifts in operational conditions and translate them into early warnings. For EHS leaders, this represents a shift from managing compliance to managing risk dynamically. The broader implication for Industry 4.0 is increasingly clear. Digital investments in automation, IoT, and video analytics have improved efficiency and visibility across Malaysian industries. However, without integration, these systems operate in silos, limiting their ability to support real-time safety outcomes. As Vision Zero targets become more central to national safety strategy, the effectiveness of these investments will depend on whether they are structured to support connected, intelligent safety systems or whether they still remain in a parallel process, disconnected from how operations actually run.

simultaneously: Is this facility safe right now, and is it running as efficiently as it should be? The architecture that makes this possible is not revolutionary in its individual components—computer vision, edge AI, and scenario-based detection have all been available for several years. What is different, is how those components are assembled and what they are pointed at. At its core, a unified safety platform deploys AI modules in existing CCTVs across a shop floor’s critical zones like production lines, loading bays, chemical storage areas, or machinery perimeters. It runs continuous scene analysis against a library of pre-trained safety and operational scenarios. It monitors every shift, every hour, flagging deviations in real time and feeding a centralised dashboard that both the safety manager and the operations director can act on. The practical significance of that

shared dashboard is harder to overstate than it first appears. When a safety event and an operational anomaly are visible in the same place, at the same time, correlations that would previously have been buried in separate reports become immediately obvious. A spike in PPE violations on Line 3 that coincides with a

One two mandates – how integrated AI safety systems connect the dots Most AI platforms in the market are built to answer one question very platform,

well. But the right one t o d a y m u s t an swe r t w o

production push on a F r i d a y

Why fraud monitoring and whistleblowing matter more than ever FORENSIC services have historically been associated with investigations— mobilised only after fraud, misconduct, or a regulatory breach had already occurred. Many such as procurement, payments, inventory, revenue, and third-party relationships. When effectively about retaliation, confidentiality, career consequences, or the perception that reports will not lead to meaningful action. An effective WB framework requires independent and For example: 0 A whistleblower reports suspicious vendor relationships monitoring

implemented, FRM enables firms to detect anomalies in near real-time; identify patterns of suspicious behaviour early; act before issues turn into material losses and strengthen confidence among regulators, investors, and stakeholders. Rather than relying on policies alone, organisations can demonstrate clear evidence of what was monitored, what was flagged, and how issues were addressed. Done well, FRM functions as a dynamic risk radar, surfacing patterns and red flags that traditional reviews may overlook in high-volume transactional data. If FRM is your radar, WB is your human intelligence network. WB remains one of the most powerful tools for uncovering fraud because it provides something data cannot: human insight. Yet many WB hotlines underperform because they are implemented as a compliance checkbox. Policies are drafted, reporting channels are introduced, and awareness materials are circulated. However, employees and other stakeholders frequently remain reluctant to speak up due to concerns

confidential reporting channels that whistleblowers genuinely feel safe using; clear triage, investigation, and escalation protocols so reports do not disappear into a black box; visible protection against retaliation, backed by consequences when retaliation occurs; and consistent leadership commitment, demonstrated through tone, resourcing, and follow-through. When people trust the system, whistleblowing becomes an early warning mechanism, not just a reporting channel. If you are looking for a whistleblowing channel, BDO EthicsLine is an independent, confidential online portal built to make reporting safe, accessible, and credible. In many cases, whistleblowing systems generate isolated signals rather than actionable intelligence. This is where FRM becomes essential. When integrated effectively, a whistleblower report becomes more than a tip—it becomes a starting point for targeted, real-time analysis.