Iowanews Headlines

As urban intelligentization evolves deeply toward AIoT, data security has become the cornerstone for the implementation and sustainable operation of artificial intelligence applications. With deep roots in high-precision artificial intelligence, AICT leverages technological innovation to address data security and privacy protection challenges in city-level intelligent systems. Its joint research achievements with universities have won top international invention awards. This honor not only recognizes the Company’s technological strength but also highlights its Physical AI development philosophy centered on data security. From the implementation of technological innovation to the construction of systematic thinking, AICT is demonstrating the profound implication of data security as the lifeline of Physical AI applications through concrete actions.

Winning International Gold Medal, Fortifying Data Security Defense via Technological Innovation

The 51st International Exhibition of Inventions of Geneva, acclaimed as the “Oscar of Inventions”, was recently held grandly, attracting over 1,000 inventions from 35 countries and regions. After rigorous evaluation, the project “Privacy-Aware On-Device Intelligent Computing Device and Applications” jointly exhibited by AICT, University of Science and Technology Beijing, and Tsinghua University stood out and was awarded the Gold Medal at the exhibition.

Founded in 1973, the International Exhibition of Inventions of Geneva is co-hosted by the Swiss Federal Government, the State and City of Geneva, and the World Intellectual Property Organization (WIPO). Held annually in Geneva, Switzerland, it is the world’s oldest, largest, and most influential international invention exhibition, focusing on showcasing scientific and technological innovations across a wide range of fields including machinery, electronics, medical care, environmental protection, energy, and information technology. The awards are judged by a panel of internationally renowned experts based on strict criteria of innovation, practicality, and commercial potential. As a core platform for global innovation competition, technological exchange, and industry-academia-research transformation, it serves as an important benchmark for measuring international invention excellence.

Currently, pervasive urban nodes such as perception cameras, radars, and roadside intelligent devices continuously generate massive data, forming a real-time portrait of urban operations. These data provide accurate real-time inputs for intelligent transportation systems, effectively monitoring traffic violations and flow, and significantly improving road safety, network efficiency, and urban management capabilities. Meanwhile, for city-level intelligent systems, data security has become a fundamental issue in urban intelligentization and poses challenges to relevant privacy regulations. AICT’s joint project “Privacy-Aware On-Device Intelligent Computing Device and Applications” precisely addresses core demands of urban intelligentization. Through the R&D and application of privacy-aware on-device intelligent computing devices, coupled with privacy-aware multimodal fused object perception and multi-target differentiated pixel rearrangement, the solution enhances the performance of high-precision AI and urban traffic management efficiency while avoiding exposure of raw sensitive data, thus fortifying the defense line of public information security.

In the R&D and application of the project, AICT and its partners carried out collaborative innovation. On one hand, they co-developed privacy-aware multi-target object perception technology, which dynamically aligns vision and radar point cloud targets through spatio-temporal consistency and maps them into a shared semantic latent space. Combined with traffic task or privacy supervision mechanisms, it reliably separates task-irrelevant targets from privacy-sensitive targets such as pedestrians.On the other hand, they jointly created multi-target differentiated pixel shuffling technology, which applies coarse-grained grids and stronger shuffling intensity to privacy-sensitive targets to destroy their semantic features, while using fine-grained grids and milder shuffling for task-relevant targets. This preserves task statistical features (e.g., violation determination) yet makes individuals visually unidentifiable, enabling back-end adjudication without exposing raw data.

Field applications show that the technology strengthens data security and privacy protection while significantly improving urban traffic management efficiency. To date, the team has been granted more than 90 patents, published over 60 high-quality papers, and received 24 international competition awards.

Dr. Jun Yan, Chairman and Chief Innovation Officer of AICT, stated that the award is a great recognition of the continuous efforts and innovative spirit of the Company’s R&D team over the years. AICT has long focused on the R&D and application of High-precision Artificial Intelligence (HAI) products. So far, its technologies and products have been widely deployed in more than 70 mega-cities including Beijing, Shanghai, Guangzhou, Shenzhen, and Tianjin, forming ultra-large-scale practical applications of high-precision AI. The award-winning project represents a major achievement of the Company’s deep engagement in data security and its effort to build a secure foundation for Physical AI applications through technology.

Reconstruct Security Thinking: Data Security Becomes the Underlying Fundamental Capability of Physical AI Applications

From the perspective of AICT, as cities enter the AIoT era, what truly determines whether artificial intelligence systems can operate in a trustworthy manner over the long term is no longer intelligence alone, but the systematic control of data boundaries, system boundaries, and capability boundaries. Data security, therefore, is not simply a feature. It is the lifeline that makes it possible for AI applications to be deployed, operated, and continuously evolved.

Artificial intelligence, the Internet of Things, and the Internet are connecting cities into a real-time “neural network”. Cameras, radar, roadside devices, parking systems, robots, and cloud platforms, all distributed across the urban landscape, continuously generate massive volumes of data and together create a real-time portrait of how a city operates. For this very reason, discussions of data security today can no longer remain at the traditional level of antivirus protection and intrusion prevention. For city-level intelligent systems, data security is becoming a more fundamental issue. Whoever can control the rules governing how data is collected, circulated, used, and handled will also hold the initiative in how the system operates. As artificial intelligence penetrates critical scenarios such as urban governance, intelligent transportation, and smart cities, data is no longer merely an input for business operations. It becomes the foundational infrastructure that determines whether a system can be controlled, audited, and sustained over time.

In the past, data security was more often understood as a technical protection issue, with the focus placed on preventing data leaks, cyberattacks, and tampering. These concerns remain important today. However, in highly connected, real-time, and deeply coordinated intelligent systems, the meaning of security has undergone a profound shift.

Once a city is digitally and intelligently restructured, data is no longer a byproduct of system operations. It becomes a core resource that supports perception, decision-making, orchestration, and governance. Whoever defines how data is collected, how it flows, how it is interpreted, and how it is used will possess a higher-order form of system control. In this sense, data security is no longer just about whether defenses can hold. It is about whether a system can uphold its baseline safeguards, retain operational initiative, and support long-term, stable functioning.

Seen from this perspective, data security is not an add-on to AI applications, nor is it a supplementary check performed before launch. It is a foundational capability that must be prioritized and strengthened throughout system development. If AI applications are to move from pilot demonstrations to deployment at scale, and from isolated use cases to mission-critical domains, the first issue that must be addressed is this fundamental prerequisite: security.

Grasp the Core of Security: Boundary Governance and Independent Controllability Are the Key Principles

In AICT’s philosophy, data controllability is the starting point of security. Data security cannot be achieved through any single isolated security product. It must be realized through clear and effective boundary governance.

From collection to transmission, from storage to invocation, and from sharing to retention, every stage of the data lifecycle must have well-defined access controls, clear circulation paths, and traceable lines of responsibility. Any data chain for which it is unclear where the data came from, where it is going, or who is using it may appear to be merely a management issue at a small scale, but in large-scale system operations it can rapidly evolve into a systemic risk.

Dr. Jun Yan believes that, truly effective data security, therefore, does not mean conducting a one-time inspection before launch, nor does it mean patching vulnerabilities only after problems arise. It means treating data as a critical asset from the very beginning of system design and governing it accordingly. Only when data boundaries are clearly defined, flow paths are properly managed, and responsibilities are explicitly assigned can AI applications be considered genuinely controllable and system operations genuinely trustworthy. For city-level public systems, data security is not merely a matter of data management. It is also a question of whether the system’s capabilities are independent and reliable. If critical capabilities are built on black boxes that are opaque, unauditable, and irreplaceable, then while the system may appear to be functioning, it is in fact difficult to maintain real control over it. Once anomalies, attacks, or supply chain risks arise, the system may quickly become passive and vulnerable. Whether system boundaries are independent and controllable therefore directly determines whether security capabilities can be effectively implemented.

This is especially true in scenarios such as intelligent transportation and smart cities, where AI systems are often directly connected to roads, vehicles, infrastructure, platforms, and management processes, affecting transportation efficiency, public service capacity, and the orderliness of urban governance. These scenarios involve wide-ranging connectivity, long coordination chains, and broad impact radii. As a result, they inherently impose higher security requirements, along with stricter demands for system independence, stability, and recoverability. Accordingly, from device onboarding and firmware upgrades to cloud services and algorithmic strategies, relevant systems must be auditable, traceable, replaceable, and capable of graceful degradation. Real security is never based on the assumption that systems will never fail. It lies in ensuring that when problems do occur, systems can still identify them quickly, isolate them in time, mitigate losses effectively, and recover smoothly. In critical domains such as intelligent transportation and smart cities, security is not an auxiliary condition. It is a prerequisite for sustained operation.

Upgrade Security Capabilities, Build a Closed-loop System, and Empower Security Governance with Physical AI

In the AI era, security capabilities must form a closed loop.This is also the direction AICT has always adhered to in its technological R&D.If security in the past relied more heavily on static rules and manual response, then in the AI era it has become a dynamic capability that must continuously evolve.Faced with constantly changing attack methods, complex abnormal operating conditions, and diverse business scenarios, systems can no longer rely solely on preset rules for passive defense. They must possess the ability to learn continuously and adapt continuously. This means that the security architecture must form a complete closed-loop.

Dr. Jun Yan emphasized that the closed-loop security system can proactively detect anomalies, rapidly assess risks, automatically generate response strategies, execute response actions, and feed the results back into models and rule systems to drive the next round of iteration and improvement.Only when such a closed-loop is established does security cease to be static or one-off. Instead, it becomes a capability that can continue to strengthen as application scenarios evolve. Truly effective security is likewise not achieved through one-time investments in equipment and manpower, but through systematic operating mechanisms that steadily suppress risk and continually accelerate response. This is also why, as AI enters the real world on a broad scale, security capability building must shift from localized protection to systemic immunity.

In AICT’s practice, AI is not only a business capability amplifier that enhances recognition efficiency and optimizes operational decisions, should also be an important tool for strengthening security capabilities. But from the standpoint of system construction, the value of AI lies not only in making systems smarter, but also in making them more secure, can be introduced into security operations themselves. Algorithms can be used for asset identification and device fingerprint management; models can be used to establish behavioral baselines and anomaly detection capabilities; automated strategies can be used to shorten response times; and feedback from outcomes can be used to continuously improve the adaptability of models and rule systems to new attack patterns, abnormal operating conditions, and complex scenarios. In this way, AI becomes not only a tool for upgrading business-side capabilities, but also a tool for enhancing security-side capabilities.

Looking ahead, Dr. Jun Yan has long maintained that data security is not an optional extra for AI applications. It is a mandatory requirement.Systems can become increasingly intelligent, connectivity can continue to expand, and applications can become ever more abundant. But all of this depends on one premise: that systems are sufficiently reliable, sufficiently controllable, and capable of withstanding the test of long-term operation. If artificial intelligence is to move toward larger-scale, deeper, and higher-quality applications, it must first safeguard this lifeline of security. This is especially true in critical fields such as intelligent transportation and smart cities, where security must be placed front and center. Only by truly establishing data boundaries, system boundaries, and capability boundaries can artificial intelligence advance steadily and sustainably in the real world.

Winning the Gold Medal at the Geneva International Exhibition of Inventions is a significant recognition for AICT in integrating data security and AI technologies. Going forward, AICT will continue to pursue disruptive innovation and cutting-edge R&D, empower all industries with HAI, and always take data security as the core lifeline of Physical AI applications.The company will keep fueling the intelligent and secure development of cities and create a better life for humanity.

Media Contact
Company Name: Artificial Intelligent Interconnection Technology Co. Ltd.
Contact Person: Wu Jinna
Email: Send Email
Country: China
Website: http://www.zhihuihutong.com