The International Civil Aviation Organisation, a specialised UN agency responsible for developing standards, oversight and monitoring, is leading the advancement of the global air traffic management system through the Global Air Navigation Plan. The introduction of automated air traffic control systems aims to improve the operational safety and efficiency of air traffic controllers. But these automated systems could introduce new risks.Around the world, air navigation service providers offer a range of services, including air traffic control. The majority of these service providers are modernising their communication, navigation, surveillance and air traffic management systems in line with the global plan. This is to accommodate the growing number of flights while improving efficiency, reducing costs and lowering aviation’s environmental impact. South Africa has been part of this global trend. The South African Air Traffic and Navigation Service has employed numerous automated systems to enhance operational efficiencies in line with global standards. During my career as an air traffic services officer in Johannesburg, I provided flight information services to aircraft, coordinated international flights with neighbouring countries, conducted high frequency radio communications with aircraft over South African oceanic airspace, assisted with search and rescue missions and participated in the replacement of various outdated air traffic control systems. Much of the work involved implementing automated technologies designed to improve air traffic controllers’ performance and ultimately increase airspace capacity and efficiency.But a noticeable increase in South African aviation incidents was reported over four years. During 2021-2022, a 69% increase in general aviation accidents and a 77% increase in incidents were reported over the previous year. Also, air traffic controllers’ incident and confidential hazard reports increased significantly in 2023-2024.These increases raised the question: if the technology was intended to improve safety, why were there more incidents?My doctoral research examined this question by exploring the effects of automation on air traffic controllers and analysing the international, regional and national regulatory frameworks governing these technologies. The findings suggest that the answer lies not in the technology alone, but in the complex interaction between air traffic controllers, automated systems and regulations.Key findings from a survey included air traffic controllers’ complacency, overreliance and the diminishing of manual skills, while automation failures increased their workload. While participants indicated that they did not trust artificial intelligence (AI) to do their work, no participant knew how AI was already employed in their operations. The legislative analyses unveiled various shortcomings at international, national and operational levels.Technology changes jobsAutomation is often presented as a way of reducing human error and workload. In air traffic management, automated systems can process large amounts of information, monitor aircraft movements and assist air traffic controllers. But introducing automation does not eliminate the human role. Instead, it changes it.Air traffic controllers are moving from actively separating traffic to monitoring systems and intervening when necessary. While this can reduce workload in some situations, it can also create new cognitive challenges. The study of human factors examines human strengths and limitations to explain why incidents occur. It has identified several risks associated with highly automated environments. Air traffic controllers may become over-trusting of automated systems, a phenomenon known as automation bias. An air traffic controller’s mental picture of what is happening on the radar and around them, termed situational awareness, can decline, resulting in increased complacency. This happens when an air traffic controller spends more time monitoring automated systems than actively controlling traffic. Skills that are rarely used may deteriorate over time.These challenges do not necessarily indicate poor performance. Instead, they reflect predictable human responses to changes in how work is organised and performed.New technology can create new risksAs part of my research, I surveyed air traffic controllers from the US, Canada, Europe, the UK, New Zealand, the United Arab Emirates and South Africa. One of the identified risks was that training programmes did not fully prepare air traffic controllers for the practical realities of operating the automated system. Procedures developed for older systems did not align with the capabilities and limitations of the automated systems. Civil aviation authorities may lack the skills to detect the slow, invisible shift in daily work away from safety procedures, known as operational drift, that occurs after implementation. Another risk is that the failure of automated systems can result in a significant increase in air traffic controllers’ workload. The timing of the increased workload matters. If it happens when air traffic controllers have developed an overreliance on these systems, they may no longer have the necessary experience to control air traffic procedurally.In some cases, the automated system performs exactly as designed, but air traffic controllers interact with it in ways that system developers did not anticipate. Electronic flight strips exemplify how air traffic controllers still favour writing on paper to record flight information, as typing could hinder their efficiency during busy times.The majority of air traffic controllers surveyed confirmed that most automated systems were not inter-operable and that they used temporary nonstandard procedures or manual interventions to bypass system limitations. In other words, they used “workarounds”. As a result, incidents occurred, such as reduced separation between aircraft, which could cause a collision.These new risks raises a further important question: who would be liable for an accident if an air traffic controller had to rely on these “workarounds” to control air traffic?Regulation has struggled to keep paceThe challenges are not limited to operational environments. They extend to international, regional and national legislators too.The International Civil Aviation Organisation recognises the rapidly growing role of automation and AI. However, it is unable to keep pace with it. Making standards is a lengthy process.Another shortcoming is that no explicit international standard requires automated air traffic control systems and AI to be certified by national regulators. Only as recently as 2023 did the European Union Aviation Safety Agency, for example, propose establishing a new regulatory framework for certifying automated systems. In South Africa, the South African Civil Aviation Authority complies with the international standards and therefore also does not require automated air traffic control systems to be certified. The authority reported a shortage of skilled technical inspectors and therefore would need automation and AI-skilled inspectors to certify systems in the future.As automation becomes more sophisticated, responsibility for safety is becoming a shared responsibility among system designers, regulators, service providers and air traffic controllers.A lesson for the age of AITechnology does not operate in isolation. Its success depends on the people who use it, the organisations that implement it and the regulatory systems that govern it. When incidents occur, the explanation is rarely as simple as blaming automation or the human operator. Interactions across the entire system shape safety outcomes.As governments and industries turn to automation and AI to solve complex problems, the aviation sector demonstrates that introducing AI is not simply a technical challenge. It is also a human, organisational and regulatory one.Aviation regulators will have to revise current legislation to determine how they will certify and oversee it, regulate AI and air traffic controller liability, and consider their role in this complex system.Cindy Hendrikse does not work for, consult, own shares in or receive funding from any company or organisation that would benefit from this article, and has disclosed no relevant affiliations beyond their academic appointment.