What Surveillance Leaders in Asia Pacific Are Really Grappling With Right Now
Over the past few weeks, I've been in Hong Kong and Singapore facilitating the 1LoD Surveillance Leaders Network sessions with senior practitioners responsible for surveillance programmes across global and regional institutions. The sessions operate under Chatham House rules. What came through is consistent with themes that have emerged from my broader work over the last quarter across trade surveillance, AI, and regulatory positioning.
I've written before about trade surveillance feeling behind the times in its use of AI compared to communications surveillance in 'Trade Surveillance and AI'. The sessions in March suggest 2026 may be the year that changes. Banks have not deployed AI, but they are certainly interested and investing. What follows is a summary of the main themes.
AI is on almost every agenda.
Artificial intelligence dominates most discussions, but the gap between interest and live deployment remains wide. In my discussions, the focus for those pushing ahead is not on replacing human judgement, but on augmenting it with better context and richer data, resulting in faster triage and greater efficiency. What's less clear is how quickly this translates into reality. Deployment, particularly in trade surveillance, remains limited, and there are very few, if any, confirmed live deployments in trade surveillance across the market.
The business case for trade surveillance is harder to make than it is for either e-communication surveillance or financial crime, where the returns are more immediate. With offshore analyst costs relatively low and AI development still expensive, headcount savings alone don't cover the investment. Internal governance poses a further challenge. Model risk management (MRM) and Responsible AI (RAI) frameworks are adding friction to the path from proof of concept to production, particularly where firms need to demonstrate how decisions are reached and why. The strong sense across both sessions is that regulators are watching and learning, but not yet directing. Very few participants had discussed AI in surveillance directly with their regulator.
Regulatory pressure is sharpening focus on fragmented areas.
Cross-product surveillance came up at participants’ request before the formal agenda began, which itself reflects how live this issue is. Firms were candid that their current strategies would struggle to detect the kinds of scenarios regulators are now focusing on, such as recent actions involving Jane Street, Bank of America, and Westpac. These cases remain too difficult to detect for most.
Trading Hub is a primary vendor offering genuine cross-product surveillance capability, but many banks use it only for fixed income, meaning coverage is partial rather than enterprise-wide. For institutions with significant India-based businesses, recent SEBI activity was a live concern. In several cases, the honest answer to whether their current setup would have detected the conduct in question was no.
The operating model debate.
Operating model discussions are particularly energetic in U.S.-led institutions, where there has historically been more variation in first and second-line ownership of surveillance activities. The APAC perspective and most global conversations I'm part of are now quite consistent. The direction of travel is toward full or majority second-line ownership of surveillance, driven primarily by conflict of interest concerns, with regulatory expectations clearly pointing firms in this direction.
Several US banks have faced or responded to regulatory pressure, including from the OCC, to move back toward second line ownership. Even where first or blended line models exist, there is increasing pressure on firms to demonstrate independent oversight.
The practical division most participants recognise is that intraday and real-time activity sits in the first line where intervention is still possible, whilst end-of-day market abuse, insider dealing and unauthorised trading reviews sit in the second line, where the conduct has already occurred and independence of review matters most.
What connects all of these conversations is a common thread: surveillance programmes are under pressure to mature in capability, in coverage, and in independence, faster than most firms had planned for. AI offers part of the answer, but the governance and operating model questions need to be resolved, fast.
A closing observation.
One theme from the networking conversations worth noting: there is a growing expectation that Hong Kong may be entering a significant growth phase. There is a growing expectation that senior expat talent currently based in Dubai and Abu Dhabi may increasingly relocate toward Hong Kong, with local institutions already expecting an influx of business and decision-making authority. Singapore, by contrast, remains constrained by visa restrictions and local-to-expat ratios that have not shifted materially since COVID.
If you're working through any of these challenges in your own organisation, I'd welcome the conversation, whether that's a specific question or just comparing notes on where things are heading.
For further reading on the governance and ethics principles underpinning effective surveillance programmes, my e-book, Behind the Screens: Understanding Employee Surveillance in Financial Services can be downloaded below.
Behind the Screens: Understanding Employee Surveillance in Financial Services
