How regulation, agentic AI and urban platforms are changing the security landscape

By Barry Norton, Fellow, Milestone Systems

As cities across Europe push ahead with their digital transformation, 2026 will be a pivotal year for video AI and smart city technologies. What was experimental just a few years ago is becoming an integral part of urban infrastructure. In Germany, the EU AI Act is shaping this development in particular. Technologies such as agentic AI, digital twins and networked urban platforms are setting new standards. At the same time, there are growing expectations that urban systems should be privacy-friendly yet closely interlinked. This is creating an environment in which video analytics are becoming a regulated, networked and increasingly intelligent part of urban ecosystems.

From contextual analysis to greater public safety

A key area of application is the real-time analysis of videos from public areas. Local authorities in Germany are moving away from surveillance-oriented solutions aimed at identification towards systems that recognise patterns, movements, people flows and anomalies. This development is in line with the legislator’s expectation that video AI should support the public interest while protecting privacy.

In practice, this leads to greater transparency, structured risk assessments and complete traceability of procurement processes. Biometric identification such as real-time facial recognition remains limited to narrowly defined exceptional cases. In all other situations, anonymisation and data protection-friendly procedures ensure that individuals do not have to be identifiable. The direction is clear: video AI focuses on what is happening in a location, while the question of who is there is deliberately pushed into the background.

Private technologies are becoming standard in urban video analysis

In parallel with regulatory requirements, technical solutions that embed anonymisation as a default setting are gaining acceptance. Providers such as the Berlin-based company brighter AI offer tools that automatically blur faces and number plates or replace them synthetically, while still enabling meaningful analysis.

In Germany, video recordings from public spaces are always processed anonymously unless there is a legally justified need for identifiable data. Data protection-friendly technologies are thus gaining importance in public tenders. Many systems analyse data directly at the camera, so that sensitive information is protected even before it is transmitted. Data protection is thus becoming a central architectural principle of urban video AI.

The smart city is growing beyond pilot projects

Germany has promoted numerous smart city pilot projects in recent years. Many of these tested urban data platforms, mobility analyses and AI-supported planning tools. A new phase will begin in 2026: cities will scale proven modules and deploy successful applications at other locations instead of setting up new, isolated test projects.

These include frequency analyses through object recognition, such as counting pedestrian and vehicle flows, measuring traffic flows and forecasting utilisation. They are based on event data such as movements and patterns rather than individual identifications. Federal programmes and guidelines promote standardised data models and interfaces. The result is a more interoperable smart city ecosystem in which video AI can be more easily integrated into existing systems for transport, planning or building management, for example.

Video AI as a central data source for urban platforms and digital twins

Modern urban platforms increasingly treat video-based analytics as one data source among many. Mobility, energy, environmental and infrastructure data flow into shared dashboards. Video AI supplements these data sets with dynamic, non-identifying information such as movement patterns, dwell times and densities.

Digital twins go one step further. They combine camera data, access controls, fire protection systems, IoT sensors and environmental data to create virtual images of entire neighbourhoods or large facilities.

This allows evacuation scenarios to be simulated, spatial changes to be tested and maintenance requirements to be identified at an early stage. In complex areas such as airports, ports or extensive campus areas, this creates a new quality in planning and operation. Simulation replaces assumptions and AI-based data strengthens modern operating and deployment processes.

Agentic AI as the pacemaker for urban security processes

Agentic AI expands the possibilities beyond classic video analysis. These systems coordinate multi-stage processes across different data sources and applications. What was initially known from software assistance systems is now finding its way into security control centres and municipal monitoring centres.

Agents check alarms, view relevant videos, compare them with access or sensor data, and bundle everything into compact case files. They suggest measures, assist with prioritisation and automate reports. This frees up control centres, allowing them to focus more on decisions rather than searching for information.

In 2026, the question will be less about whether agentic AI is useful and more about which recurring processes are best suited to it.

Wearables and AR are evolving into intelligent operational aids

For emergency services, AI-supported wearables and augmented reality glasses are becoming more suitable for everyday use. Passive devices such as bodycams are becoming active assistants that classify situations, provide context and can be addressed using natural language.

Employees can ask when an area was last inspected or receive information via voice command without having to look away. The growing demand for AR-supported training, remote support and operational assistance systems shows that this technology is finding its place in urban processes in the long term.

Governance and public dialogue as the basis for acceptance

The more AI is integrated into urban infrastructure, the more important its responsible use becomes. Cities are developing guidelines that define how video AI is used, how sensitive data is protected and where clear boundaries lie. Critical alarms continue to be reviewed by humans. At the same time, local authorities are communicating more openly about where video AI is used and how anonymisation works. This strengthens trust and social acceptance.

A new operational reality for smart cities

The interplay of context-sensitive video analysis, agentic AI, digital twins and intelligent wearables shows that smart cities in Germany will enter an operational phase in 2026. The focus is on integration and responsible use. Systems are becoming more networked, more data protection-oriented and better adapted to regulatory requirements. AI is becoming a tool in everyday life and smart city solutions are becoming scalable.

In this new phase, cities and operators are benefiting from clear data architectures, simulation-based planning and human oversight. Video AI is becoming a building block of a modern, responsive and intelligently controlled urban environment.