By Ben Parker, Co-founder and CEO, eflow

On Friday 3rd July 2026, the Market Abuse Regulation (MAR) marked its 10-year anniversary. A decade on, MAR has undoubtedly raised the bar for market conduct – but it has also exposed how difficult it is for firms to consistently detect and evidence market abuse in increasingly complex, data-heavy markets. Despite a decade of regulatory evolution, market abuse remains a persistent issue.

Think about the level of change we’ve had since 2016. The Brexit vote had only just taken place, digital assets then took off, triggering the need for new regulations, a pandemic happened, driving everyone online, and then there was the AI boom. All pretty substantial changes, each one significantly impacting how traders trade and, crucially, how market abuse can occur. Newer trends now include 24/7 trading and prediction markets.

In this decade of profound change, what has MAR achieved? And what structural challenges remain or have emerged?

MAR shifted the problem from regulation design to surveillance effectiveness

MAR was an EU regulation, so when the UK left the EU, it eventually became onshored into UK law in 2020. It was an important step in addressing a sector-wide problem for how regulation could be designed to tackle market abuse. MAR established specific definitions for market abuse offences like inside information and market manipulation, what entities were required to disclose information and how to prevent and detect abuse.

But by setting out these criteria, the problem has since become how to surveil for these forms of abuse effectively and how to establish quality data that enables firms to do this. Firms are obligated to identify and report potential cases of market abuse to the relevant regulatory authority under MAR. But, as outlined before, the trading landscape has become increasingly digitised and complex, with a wide range of financial instruments, players, assets, technologies and techniques. Consequently, firms still face an ongoing, and increasingly complex, challenge to identify market abuse across fragmented trading environments.

Adapting to new challenges

Once regulation is implemented, it needs to adapt regularly to changes in the market. A trend we’ve observed from our own research is a growing focus by regulators on deficiencies in data, systems and controls. Last year, for example, the majority (59%) of global enforcement action relating to market abuse focused on systems and control failures.

MAR has successfully set a baseline for how firms can detect and report abuse. Yet within this structure, there is a gap between the expectation for strong trade surveillance foundations like comprehensive risk assessments and data governance and the reality of firms’ systems and processes. Firms are aware of the issue, solving it is just more challenging.

Risk is constantly evolving and firms need to continuously update their risk frameworks while ensuring their trade surveillance platforms can be configured dynamically for different asset classes. In particular, the need for dynamic alert thresholds has come on hugely since MAR was first introduced – they’re now integral for systems navigating a wide-ranging trading environment where abuse can take place across different channels.

The second key enforcement transition is a focus on evidence over intent – how can firms evidence their capability and explain the reasoning behind alerts? I wrote about the importance of ‘regulatory explainability’ in April, where I outlined how regulators are no longer accepting “we have a trade surveillance system in place” as a valid explanation from firms – they need to demonstrate how that system is helping them remain compliant and spot market abuse. And underscoring these changes is, of course, AI.

Where AI is helping – and where it is not yet solving core surveillance challenges

Over the last year and a half, AI has evolved from a trending buzzword to a practical tool that some firms are actively deploying as part of their regulatory strategy. Its main benefit in the context of trade surveillance is by helping to reduce false positives and triage alerts, which can be a substantial problem for compliance teams, through dynamic thresholds. The application of AI also enables firms to build out more sophisticated surveillance frameworks while maintaining a similar size compliance team. Yet this is where considerations around explainability, governance and human oversight are critical.

Compliance teams and firms, not the technology, are accountable for regulatory decisions and enforcement action will be levied against them for any errors – this is why human oversight is so important for monitoring outputs and system accuracy. Concurrently, when firms are using AI systems, they must ensure they are explainable to match expectations. That involves using a platform where they can track alerts or insights to their source data to show how an output was reached, over the use of opaque ‘black box’ systems.

The main point is that AI can help to improve operational efficiency and identify abuse faster, but it can’t replace the underlying foundations of robust governance, human judgement and specialist expertise.

Strong foundations crucial for the next decade

The challenge for regulators is always in keeping pace with market and technological changes. Abuse typologies largely remain the same. What changes is how they can take place, whether that’s through digital assets or AI-powered trading, where, and how quickly. Prediction markets, for example, are a relatively new trading space, and they’re growing fast. They are highly susceptible to insider trading – the abuse type is the same, but it’s happening in a new way and with greater ease.

MAR has been great at forming robust, clear frameworks for market abuse and detecting it. The challenges emerge in the gap between regulatory expectations and the reality of firms’ capabilities to meet them in a market that constantly seems to move one step ahead. So, MAR needs to incorporate these new trends and help firms establish the necessary controls to monitor these channels for abuse. But in doing this, what’s crucial is that the foundations of governance and data quality remain. For example, if data is consolidated and dynamic thresholds are in place, then it is much easier to re-calibrate systems to monitor for abuse in new markets or techniques.

The next decade will see even greater change to market abuse regulation and surveillance capabilities. Ensuring regulation can keep up with trends, and firms’ capabilities can keep up with these expectations, will be the critical task.