Since the beginning of this year, intelligent agents have become an important development direction in the field of artificial intelligence. An intelligent agent refers to an agent that can perceive the environment and take action to achieve specific goals. It can be software, hardware, or a system with autonomy, adaptability, and interaction capabilities. Intelligent agents are widely used in the field of artificial intelligence, commonly in automation systems, robots, virtual assistants, and game characters. Their core lies in the ability to learn autonomously and continuously evolve to better complete tasks and adapt to complex environments. At the 13th Internet Security Conference (ISC. AI 2025) held recently, agents became the focus of hot discussion. Experts attending the meeting believe that as the development of artificial intelligence enters the second half, intelligent agent technology is accelerating its breakthrough beyond laboratory boundaries, deeply integrating into core industrial fields, and increasingly becoming a key engine driving industrial transformation. AI continues to evolve. According to the data of China Academy of Information and Communication, there are more than 35000 global AI enterprises and 5100 Chinese AI enterprises, accounting for about 15% of the world's total. The scale of the industry continues to grow, forming a complete industrial system covering infrastructure and industrial applications. According to the 56th Statistical Report on the Development of Internet in China released by China Internet Network Information Center, the scale of China's AI industry will exceed 700 billion yuan in 2024, maintaining a growth rate of more than 20% for consecutive years. In the first half of 2025, generative artificial intelligence products have achieved comprehensive progress from technology to application, with a rapid increase in product quantity and continuous expansion of application scenarios. From the perspective of users, as of June, the proportion of users using generative artificial intelligence products to answer questions was the highest, reaching 80.9%. "The iteration of artificial intelligence technology has accelerated, and the innovation in the field of artificial intelligence is showing a trend of group breakthrough. The deep integration of artificial intelligence and the real economy is more obvious." Zhao Zhiguo, executive vice director of the expert advisory committee of the China Internet Association, said. With the continuous development and innovation in the field of artificial intelligence in China, a number of high-quality domestic artificial intelligence products have accelerated to go to sea, and China's influence in the field of artificial intelligence has significantly increased. For example, within less than 20 days of its launch, DeepSeek has surpassed 30 million daily active users worldwide, becoming the fastest-growing generative artificial intelligence application in terms of user growth. In addition, China has become the world's largest holder of artificial intelligence patents, accounting for 60%. Jiang Hongchao, Deputy Director of the Beijing Municipal Bureau of Economy and Information Technology, introduced that currently, artificial intelligence technology represented by large models is developing rapidly, injecting strong momentum into the digital transformation and upgrading of the real economy. Beijing has gathered over 40% of the country's top artificial intelligence talents, with over 2400 artificial intelligence enterprises and a core industry scale of nearly 350 billion yuan. As of now, a total of 132 large-scale models have been registered and launched, accounting for over 30% of the total nationwide, firmly ranking first in the country. Zhou Hongyi, founder of 360 Group, believes that the ability of big models is becoming stronger and stronger, but having only big models is not enough. The two major pain points of large models in enterprise applications in the past were insufficient reasoning ability and lack of independent work ability. The former has significantly improved in the past year, but the latter has not yet been resolved. The large model lacks "hands and feet", cannot use tools, and cannot work directly. If AI development stays at the stage of large-scale models, it can only become a toy rather than a productivity tool. Wu Hequan, honorary chairman and professor of the 13th Internet Security Conference, believes that "agent as a service" has become the core symbol of Internet evolution. "User led operation" will turn to "agent independent execution+human supervision and decision-making", and people will transform from operators to AI behavior decision makers. The evolution of large models to intelligent agents is inevitable, and intelligent agents are the solution. It can understand goals, plan tasks, call tools, have memory, and achieve complete delivery from requirements to results. Its core lies in the ability to use tools and utilize large models for complex task planning. ”Zhou Hongyi said. The increasing network risks and the emergence of intelligent agent hackers have exacerbated the asymmetry of network attack and defense. Artificial intelligence is both a risk and an opportunity for network security. Zhao Zhiguo stated that the increasingly close integration of artificial intelligence and security fields has not only given rise to more proactive and adaptive technological tools for network security, but also promoted the reconstruction of the security system from passive defense to intelligent driving, and its "security capability multiplier" effect is rapidly emerging. Zhao Zhiguo analyzed that the security risks behind technological innovation are intertwined and overlapping. Specifically, firstly, technological fragility brings about endogenous security risks. At present, the innovation of artificial intelligence technology is driven by three elements: data, algorithms, and computing power. The inherent attributes of data dependence, algorithm complexity, and computing power concentration make the unique security risks of model illusion, algorithm black box, and data trustworthiness more prominent. Secondly, malicious exploitation amplifies security crises. The attack methods driven by artificial intelligence are developing towards intelligence and automation, significantly reducing the attack threshold and increasing the attack frequency, while further enhancing the concealment and variability of attacks. New types of network threats such as deepfake fraud, automated attacks, and open-source ecosystem poisoning are increasing day by day. Thirdly, the complexity of governance is becoming increasingly prominent. The widespread use of artificial intelligence technology, application promotion in various industries and fields, and deep penetration in multiple scenarios have made its security governance involve multiple dimensions such as technical standards, laws and regulations, application security, and ethical norms. The speed of regulatory framework improvement still lags behind the speed of technological development and application, and the difficulty and uncertainty of governance continue to rise. Zhou Hongyi believes that current enterprises are facing dual challenges. On the one hand, security operation experts are scarce and have a long training period; On the other hand, the emergence of "intelligent agent hackers" has brought attack and defense into a new stage of "machine to machine" confrontation. A single hacker can manipulate hundreds of intelligent agent hackers to launch automated attacks in batches, exacerbating the risks of cyber warfare. Hao Zhiqiang, Director of the Education and Examination Center of the Ministry of Industry and Information Technology, stated that as large models evolve towards embodied intelligence, intelligent agent technology is reconstructing the security attack and defense system. Transitioning from passive defense to active immunity, upgrading from single point protection to global collaboration. This profound transformation urgently requires the construction of a modern capability system that supports the construction of a digital China, with talent as the foundation, innovation as the driving force, and security as the cornerstone. Ensuring safety and reliability to promote the integration and development of AI technology and digital security has become the key to assisting digital construction and accelerating the development of the digital economy. Ren Xianliang, Vice Chairman of the Social Construction Committee of the 13th National People's Congress and Secretary General of the World Internet Conference, said that while giving full play to the potential of AI, we should ensure the security, reliability, controllability and fairness of its development, and make it an international public product that protects security and benefits mankind. Ren Xianliang believes that there is imbalance in the development of global digital technology at present. Some countries and regions are weak in digital infrastructure construction and lagging in capacity building, which not only makes it difficult to enjoy the benefits of Internet development, but also exposes greater vulnerability in the face of network security attacks. All parties should embrace the opportunities of artificial intelligence development, actively participate in the development and governance of artificial intelligence, and promote balanced development of global digital infrastructure through open cooperation. By utilizing technology transfer, capacity building, and institutional coordination, we can enhance the level of network security protection for developing countries and regions, and ensure network security through digital inclusiveness. Wu Hequan suggested that we should deepen the collaboration of industry models and solve the "last mile" problem; Accelerate the innovation and evolution of large-scale model technology, improve efficiency and accessibility; Innovatively upgrading the digital security protection system and building an AI driven proactive immunity; Deepen international cooperation and jointly build an open, inclusive, and secure digital future. Faced with the increasingly severe and complex threat situation, traditional passive and fragmented security protection is no longer sufficient, and fundamental changes are urgently needed. Security capabilities must accelerate their migration to the cloud and service-oriented transformation, achieving elastic scalability and on-demand delivery, making Security as a Service a solid barrier to safeguard the digital world. Digital security enterprises should focus on building a systematic security operation framework that covers the entire lifecycle of prediction, defense, detection, response, and recovery, integrating processes, personnel, and technology to form a closed-loop capability. More importantly, we need to fully utilize the powerful capabilities of AI, especially large models, to empower threat hunting, anomaly detection, and automated response, and promote the leap of security defense from "remedial measures after the fact" to an active immune mode of "early warning and blocking during the event". The integration of artificial intelligence and security fields is becoming increasingly close, and we must adhere to technological innovation to build a solid foundation for safe development; Adhere to integration empowerment and strengthen the ecological development of industries; Adhere to collaborative governance and form a joint force for safe development. ”Zhao Zhiguo said. Zhou Hongyi introduced that the security intelligent agent launched by 360 is based on the "brain" of the security big model, adding "hands and feet" to it, especially the L4 multi-agent swarm, which can replicate the abilities of advanced human security experts. 360 is promoting the intelligentization of its entire product line, committed to shaping digital security with safety intelligence. Security is the foundation of digitization, and AI is the pinnacle of digitization. 360 will continue to adhere to the dual development of 'security+AI', guarding the era with security and defining the future with AI. ”Zhou Hongyi said. (New Society)