Predictive taxpayers and the future of revenue

There is a growing divide in the world of taxation, but it isn’t just about who pays the most. It is about who knows the future. For many years, taxation has followed a simple and familiar rhythm. People and businesses earn income, keep records, calculate profits, and then pay taxes based on what has already happened. Governments design tax laws around past income and completed transactions. Audits, penalties, and revenue forecasts are all built on this backward-looking system. It has given taxpayers clarity and the state a sense of control.

Today, that rhythm is slowly changing. The change is not loud or dramatic, but it is powerful. Artificial intelligence is quietly reshaping how large businesses think about tax. Instead of waiting for rules to be enforced or audits to begin, many companies are now trying to see the future before it arrives.

Across the world, large corporations are using advanced AI tools to predict how tax authorities might behave tomorrow. These systems study past audits, court cases, policy speeches, budget signals, political trends, and even public opinion. They estimate how likely it is that a certain transaction will be questioned, how strict enforcement might become, or how laws may be interpreted in the near future. Based on these predictions, companies adjust their decisions today. This has given rise to a new figure in modern taxation: the predictive taxpayer.

A predictive taxpayer does not simply react to tax rules. Instead, it tries to stay one step ahead. Using powerful algorithms, companies forecast possible tax outcomes before income is earned or laws are changed. Decisions about investments, pricing, profit sharing, and business structure are increasingly guided by what AI systems believe might happen next, not just by what the law says now. Tax planning has become less about reading legislation and more about predicting behaviour.

This marks a deep change in how tax compliance works. Traditionally, compliance meant following written rules and responding when the tax authority acted. Now, compliance is shaped by expectations. A company may choose to pay more tax today or disclose information early because its system predicts tougher enforcement in the future. At the same time, if enforcement is expected to be weak, delayed, or under-resourced, more aggressive tax positions may be quietly accepted as manageable risks. In this world, tax behaviour is guided by probability, not principle.

From the Government’s point of view, this shift brings both opportunity and danger. On the positive side, predictive behaviour can sometimes improve short-term revenue collection. If companies expect stricter audits, they may avoid disputes, settle early, and take safer positions. This can reduce court cases, speed up collections, and ease the administrative burden on tax officers. In the short run, revenue flows may even appear more stable.

However, the long-term risks to revenue generation are far more serious. Predictive taxpayers can move faster than governments. When policy discussions hint at future reforms such as new digital taxes, environmental levies, or changes to profit-shifting rules, AI systems may immediately recommend restructuring. Assets can be relocated, supply chains redesigned, and profits shifted before laws are passed. By the time reforms take effect, the tax base may already have changed. Revenue is not lost through illegal evasion, but through legal anticipation.

This kind of revenue loss is difficult to detect. There is no obvious wrongdoing. There are no dramatic scandals. Yet the result is the same: governments collect less than expected. This is especially dangerous for countries that rely heavily on domestic revenue to fund education, healthcare, infrastructure, and social protection.

Another major concern is fairness. Predictive tax tools are expensive and complex. They are used mainly by large multinational companies with access to data scientists, global advisors, and advanced technology. Small and medium-sized businesses, and ordinary individual taxpayers, do not have this advantage. They follow the rules as they are written and react when enforcement occurs. Over time, this creates a two-speed tax system, one for those who can predict the future, and another for those who cannot.

When people believe that the tax system favours the powerful, trust begins to erode. Tax compliance is not built only on fear of penalties. It also depends on a shared belief that the system is fair. If that belief weakens, voluntary compliance declines. This is a serious risk for long-term revenue mobilisation, especially in developing economies.

Revenue forecasting also becomes more complex in a predictive world. Governments traditionally rely on historical data to estimate future collections. But when taxpayers constantly change behaviour based on future expectations, past trends become unreliable. Revenue becomes sensitive to perceptions, rumours, and signals, not just economic performance. Tax administration slowly shifts from rule enforcement to risk management. This requires new skills, new tools, and a new mindset within revenue authorities.

Sri Lanka is not outside this global shift. While the country’s tax administration is still largely reactive, many multinational companies operating locally already use predictive systems designed overseas. These systems do not respect borders. If tax authorities do not develop similar analytical capacity, a widening gap will emerge between how taxpayers plan and how the state responds. That gap represents a real risk to future revenue, especially at a time when Sri Lanka needs strong and stable domestic revenue to support recovery and development.

The key question, then, is whether predictive taxpayers pose a revenue risk for governments. The answer is yes but only if governments remain passive. Predictive behaviour itself is not the enemy. The real danger lies in uneven capability. When only taxpayers can see ahead, the system tilts in their favour.

The solution is not to resist technology or attempt to ban prediction. That would be unrealistic and counterproductive. Instead, governments must learn to govern prediction. Tax authorities need their own predictive tools to identify risks early, understand behavioural patterns, and anticipate avoidance strategies before they spread. This is not about copying corporate systems, but about building institutional intelligence.

Clear and timely laws also matter more than ever. When rules are vague or delayed, predictive systems thrive on uncertainty. Faster reforms, clearer drafting, and consistent enforcement reduce the space for speculative planning. Transparency about enforcement priorities can also help. When taxpayers understand what will be enforced and why, prediction can support compliance rather than avoidance.

International cooperation is equally important. Predictive tax planning easily crosses borders. Without data sharing, coordinated standards, and ethical guidelines for AI use, countries with weaker institutions risk falling behind. In the future, revenue loss may occur not because taxpayers break the law, but because governments fail to anticipate behaviour.

The predictive taxpayer is not automatically a threat. Used responsibly, predictive tools can reduce disputes, improve compliance, and help governments plan better. But without strong institutions, clear rules, and adaptive governance, prediction can quietly weaken the meaning of tax obligation and undermine long-term revenue stability.

In the age of artificial intelligence, taxation will no longer wait for events to happen. The challenge for Sri Lanka and for the world is to ensure that foresight strengthens the tax system rather than reshaping it in favour of those who can see the future first.