Video: 中国の国家戦略から読み解く最新脅威と、組織が講ずべき対策 | Duration: 4908s | Summary: 中国の国家戦略から読み解く最新脅威と、組織が講ずべき対策 | Chapters: Introduction to China's Threats (142.47s), Threat Spectrum Model (275.485s), Policy Shifts (419.68s), Strategic Technology Priorities (593.11s), Taiwan Strategic Tensions (783.65s), Influence Operations (937.995s), Targeting and Recruitment (1069.88s), Infrastructure Scale Growth (1275.885s), Vulnerability Discovery System (1605.465s), State Cyber Ecosystem (1952.715s), Attack Infrastructure Tactics (2442s), Red November Targeting (2849.105s), Practical Defense Strategies (3150.705s), Q&A and Closing (3565.825s), Japan APT Threats (3828.795s), Third Party Risk (4012.76s), Cloud Targeting Trends (4253.865s), Chinese Intelligence Targeting (4504.83s), Defense Priorities (4691.23s)
Transcript for "中国の国家戦略から読み解く最新脅威と、組織が講ずべき対策": Thank you. So good afternoon. Thank you for joining us. Today, I'll explain how China's strategic objectives are shaping the threat landscape. My main point is simple. Chinese threat activity is easier to understand when we connect operations to policy goals. Cyber operations are one part of that picture. Beijing also uses diplomacy, investment, regulation, research networks, commercial partners, and influence activity. I will address four questions. What does China want? Who helps advance those goals? Which organizations face greater exposure? And what can defenders do now? I will focus on patterns, not simple predictions. One incident does not prove a national strategy. We will look for repeated links among policy priorities, targeting, timing, infrastructure, and behavior. This slide gives us a simple way to organize the threat. It is a working model. It is not a fixed ranking. On the left are over influence operations. They are public and often produce a large amount of content. Each item may have limited direct impact. In the middle are covert influence networks. They hide their identity or their sponsor. Their scale and impact can vary. On the right are espionage and other hybrid threats. These activities may be less visible. Their consequences can be much greater. China can use several parts of this spectrum at the same time. Official policy and investment can support long term goals. Covert networks can shape information or collect intelligence, and cyber operations can provide access to systems and data. This is why we should not begin with malware. We should begin with the objective. One source note is necessary before we use these slides. They summarize the October 2025 recommendations for the fifteenth five year plan. They do not reproduce the final plan adopted in March 2026. The recommendations still show the policy direction clearly. The final plan retained the broad emphasis on security, innovation, industrial development, and high quality growth. This slide shows five important shifts. First, development and security are now treated together. Second, technological self sufficiency has a higher priority. It is more closely connected to industrial modernization. Third, implementation should reflect local conditions. Fourth, domestic consumption receives more attention. Fifth, China continues to seek foreign investment and foreign talent. These goals can operate at the same time. China can reduce dependence in strategic areas while continuing to attract foreign capital, technology, and expertise. For risk teams, this combination matters. It can create commercial opportunities. It can also increase pressure around sensitive knowledge, technology, and talent. Now this slide is a map of strategic technology priorities. Please read the two columns as separate lists. The row rows are not direct replacements. The recommendations name new energy, advanced materials, aviation, aerospace, and quantum technology. They also name integrated circuits, industrial machine tools, basic software, and six g communications. Other priorities include embodied artificial intelligence, bio manufacturing, brain computer interfaces, hydrogen, and nuclear fusion power. These priorities do not prove that every company in these fields will be be targeted. They do show where technical knowledge has strategic value for Beijing. The exposure extends beyond large manufacturers. It can include universities, laboratories, startups, investors, suppliers, professional services firms, and individual researchers. Organizations in these fields should protect research data, specialist talent, supply chain information, and business partnerships. They should also watch for unusual recruitment, consulting investment, and collaboration approaches. The strategic objective helps us define the likely collection requirement. The collection requirement helps us identify who may face greater risk. Taiwan remains central to China's strategic planning. The recommendations still state a preference for peaceful cross strait development. However, the language is firmer than in the recommendations for the fourteenth plan. The newer text gives greater prominence to fighting separatism and exposing opposing external interference. The two columns are lists in the order of mention. They are not direct word for word replacements. Recorded future assessed that the language almost certainly reflects a harder position. That is an analytic judgment about policy direction. It is not a prediction that conflict is inevitable. For organizations in Japan, Taiwan related tension can affect several areas. These include cyber activity, semiconductor supply chains, maritime transport, financial markets, and the information environment. The risk is not limited to organizations located in Taiwan. It can also affect companies that support regional communications, logistics, defense technology, and critical infrastructure. We We now move from strategic goals to the networks that can support them. Cyber espionage is not the only way to advance a strategic objective. This slide shows an influence network. Recorded future tracks this activity as Empire Dragon. It is also commonly known as spamouflage. We assess a link to China's Ministry of Public Security. The network conducts coordinated inauthentic behavior. Multiple accounts or false identities work together. They try to make organized messaging appear independent or authentic. These examples address political and security issues from across the Indo Pacific. They include Tibet, India, Myanmar, Japan, Taiwan, The Philippines, and the South China Sea. The the logos show artificial intelligence tools used to generate content. They do not suggest that the technology companies supported the operation. Artificial intelligence can reduce the cost of producing images and text. It can also help operators adapt content for different audiences. However, more content does not automatically produce more influence. Distribution, credibility, timing, and trusted messengers still matter. For organizations, the risk includes false narratives, impersonation, and content that exploits real crisis. Monitoring should focus on coordinated behavior, not only on whether one item is false. This slide shows a different form of targeting. It does not depend on malware. The example on the left is an online recruitment notice. It sought people with policy expertise and government connections. It offered short term remote work, and it also required a confidentiality agreement. The example on the upper right comes from public reporting. A staff member on the United States House China Committee was reportedly offered $10,000 for policy insights. The reporting raised questions about possible intelligence collection for China. The activity also used professional looking policy websites. These sites can make an approach appear legitimate. We should be careful with attribution. A false profile is not automatically part of a state operation. We should look for several links across identities, infrastructure, content, and behavior. For organizations in Japan, the lesson lesson is practical. Recruitment messages, consulting requests, conference invitations, and research partnerships can all carry collection risk. This is especially true when they seek nonpublic information or access to sensitive people. Security awareness should therefore include executives, researchers, policy teams, and public affairs staff. The threat does not stop at the security operations center. Security This slide shows the scale and growth of related infrastructure. The line chart tracks cumulative domain registrations. Growth began to accelerate after 2023. It increased sharply during 2025. We assess that geopolitical events and easier access to artificial intelligence tools probably contributed to this growth. This is an assessment. It is not proof of one single cause. The activity is global. Much of the observed targeting focused on The United States. We also saw broad exposure across the Asia Pacific region. The chart on the lower right shows where websites were hosted. It does not show where victims were located. The data showed 61 sites hosted in Singapore. It showed 55 in Hong Kong. Japan appeared with three sites in this dataset. Hosting location does not establish government involvement. It shows where the infrastructure was placed. Organizations should monitor the full network. They should not monitor infrastructure only in China. This case shows how one network can support several objectives. The infrastructure included false leak websites targeting the Philippine government. The site also contained inauthentic forum posts and comments. These created the appearance of real public engagement. Some narratives overlapped with content promoted by covert Chinese influence networks. That overlap alone does not prove direct government control. However, the combination is important. The same infrastructure can collect information, create false credibility, and amplify political narratives. Defenders should examine the complete network, not each website or account in isolation. Now a zero day is more than a technical vulnerability. It can also be a strategic asset. Technical teams ask whether a flaw exists and whether it can be exploited. Governments ask different questions. Could this vulnerability provide access to an adversary, and how much damage could it cause if another state used it first? This is why vulnerability discovery matters to national security. The value comes from finding the flaw early and deciding how to use that advantage. China's vulnerability system helps create that opportunity. Influence in virtual targeting show one part of the wider system. We will now return to technical capability. China has built a large system around vulnerability discovery. This system has four main parts. The first part is mandatory reporting. Covered technology providers must report security flaws to the Ministry of Industry and Information Technology. The second part is incentives. Researchers and member organizations can receive recognition and financial rewards. The third is talent development. Competitions and academic programs help identify skilled researchers. The fourth is private sector participation. Technology and security firms contribute research and technical capability. Together, these parts create large pipeline for discovering vulnerabilities. The same system can support defensive goals, commercial research, and national security objectives. We should not assume that every researcher or company is conducting offensive work. The important point is that the state can draw from a large pool of expertise. That scale creates more opportunities to discover a valuable flaw early. Now this slide shows the wider pipeline. Vulnerability reports can come from companies, universities, competitions, and private researchers. They can enter several government linked channels. These channels include the Ministry of Industry and Information Technology. They also include China's National Computer Emergency Response Organization and the Ministry of State Security. Some vulnerabilities are disclosed so that vendors can develop patches. Others may be retained and assessed for offensive use. We do not know how many vulnerabilities follow each path. We should not assume that every reported flaw becomes an offensive tool. However, a larger research network gives the state more opportunities to find a useful flaw first. The strategic advantage is early access and choice. A contractor may provide one capability, a university may provide research. An intelligence service may define the requirements, and an operational group may use the results. These boundaries between roles are not always visible from the victim network. One part of that system is military. In April 2024, China reorganized the People's Liberation Army. Two organizations on this slide are especially relevant. The information support force supports the security and continuity of military networks. It also supports data systems and command infrastructure. The cyberspace force has offensive and defensive cyber responsibilities. Both sit within a military structure directed by the Central Military Commission. This shows that cyber capability is part of military planning. It is not separate from broader operational goals. However, the military is only one part of the system. The Ministry of State Security has foreign intelligence and counterintelligence responsibilities, and the Ministry of Public Security also has relevant cyber and influence functions. Universities, research institutions, security companies, contractors, and individual specialists can also contribute capability. The result is a broad state and state aligned ecosystem. China also acquires cyber capability through procurement. The People's Liberation Army, the Ministry of State Security, and the Ministry of Public Security all purchase technical services. The examples include cyber training platforms, mobile device exploitation tools, and equipment for close range wireless attacks. This allows government organizations to draw capability from private companies and specialist contractors. The relationship may involve research, software development, infrastructure, training, or operational support. This procurement system helps explain why state aligned activity can be difficult to attribute to one agency or unit. Red foxtrot provides a clear example of attribution to the People's Liberation Army. Recorded future linked the group to PLA unit six nine zero one zero. That unit belongs to the Western Theater Command. The assessment used several forms of evidence. These included operational security mistakes, procurement records, infrastructure, and specific malware families. The group targeted government, defense, and telecommunication organizations across Central Asia. This example shows why attribution requires more than one technical indicator. It requires intelligence from several independent sources. We can now see the scale of observed activity. From January 2025 through June 2026, Recorded Future tracked more than 50 Chinese threat groups. We observed targeting in more than 50 countries. Four groups were associated with the People's Liberation Army, four were associated with the Ministry of State Security, and most were not publicly tied to one specific Chinese agency. The focus included critical sectors in Taiwan, Southeast Asia, Japan, and The United States. Government entities in Europe and the global South also faced activity. Recorded future also tracked more than 12 in automation networks. These networks used compromised devices or leased systems to hide activity. The map does not mean that every organization in a highlighted country was compromised. It shows where recorded future observed targeting during this period. The main lesson is breadth. The activity is global, but it follows clear strategic and sector priorities. Several technical trends appear across recent Chinese operations. First, actors increasingly exploit Internet facing devices. These include routers, firewalls, virtual private networks, and email systems. Second, they use compromised devices and relay networks to hide their activity. Third, they use legitimate services such as cloud storage and content delivery networks. This helps malicious traffic blend with normal traffic. Finally, different groups contribute to share malware, infrastructure, and vulnerability research. This suggests that specialist suppliers support more than one operational group. The result is an ecosystem that is larger than any single actor. Chinese actors often hide behind relay networks. Some relay networks use compromised home and office devices. The owners may not know the attackers are using their equipment. Other relayed networks use leased virtual servers. The operator controls these systems directly. Mixed networks use both compromised devices and leased servers. This creates several layers between the operator and the victim. The visible source address may therefore be misleading. It may belong to an ordinary router, a small office, or commercial hosting provider. This makes the real operator harder to identify. It can also make simple country blocking less effective. Defenders should examine behavior, infrastructure relationships, and changes over time. One Internet address is rarely enough for confident attribution. The next example shows how hidden access appears in a real campaign. It's Niva. Recorded future tracks this actor as Red Mike. It is also known as Salt Typhoon. Between December 2024 and January 2025, the attack actor attempted to exploit more than 1,000 unpatched Cisco network devices. Recorded future observed compromised telecommunications providers in several regions. The affected countries included The United States, Italy, South Africa, Thailand, Saudi Arabia, and Kenya. The actor used known vulnerabilities in Internet facing devices. It separated scanning infrastructure from post compromise in con infrastructure, and it used tunnels to move data. These choices supported quiet and persistent access. They also made detection and attribution more difficult. The campaign illustrates three broader trends. Chinese actors continue to target telecommunications providers. They often exploit known flaws and expose devices, and they value access that can remain useful over time. For defenders, the practical message is direct. Inventory Internet facing devices, patched known exploited vulnerabilities quickly, and watched for configuration changes in new tunnels and unusual outbound traffic. Now Red November shows a different pattern. This group has been active since at least 2024. Recorded future observed targeting of government, aerospace, defense, telecommunications, information technology, manufacturing, and religious organizations. The activity had global reach. The strongest focus was in The United States. We also observed activity in Europe, The United Kingdom, South America, Africa, and Central And South Asia. The colors on the map have different meanings. Red shows observed victim activity. Blue shows scanning or reconnaissance. Scanning does not mean that compromise occurred. The group often examined organizations connected to current geopolitical issues. This does not prove that every scan had direct government tasking, but it does show that geopolitical context helps explain changes in target selection. The two case studies show why organizations should not rely only on actor names. Targeting can change as the policy needs change. Retinuver member combines several types of tools. For initial access, the group targets exposed email systems, firewalls, virtual private networks, and collaboration platforms. Sorry. I think I have I have lost names there. Oops. Strategic targeting can also follow economic priorities. This slide compares Chinese cyber activity against Indonesia with major mining investments and commercial agreements. Indonesia mining Indonesia has strategically important reserves of nickel and other materials. These resources support batteries, advanced manufacturing, and the energy transition. The timing does not prove that every cyber operation was connected to a mining deal. Correlation is not the same as causation. However, it shows why economic context matters. Organizations that hold critical resources, technology, or commercial information may become intelligence targets. We can now turn from the threat to practical defense. First, assume that an attacker may already have access. Look for unusual behavior inside the network and do not depend on only blocking the first point of entry. Second, hardened systems against known vulnerabilities. Zero days receive attention, but many campaigns still use known flaws. Prioritize vulnerabilities that are actively exploited and exposed to the Internet. Third, improved detection of persistent access. Monitor Internet facing devices, identity systems, administrative tools, and unusual outbound connections. Fourth, secure identities and privilege access. Use strong authentication, limit standing privileges, and review access that is no longer required. Fifth, connect technical priorities to strategic exposure. A semiconductor company and a telecommunications provider may face different collection requirements. Their intelligence needs should reflect that difference. The goal is not to predict every exploit. The goal is to reduce the attacker's options before and after initial access. Taiwan reported another increase in cyber activity against critical infrastructure in 2025. The largest reported increases affected energy organizations and hospitals. Communication and transportation also remained exposed. According to the Taiwan government report, exploitation of hardware and software vulnerabilities was the most common technique. Distributed denial of service attacks and social engineering were also significant. These figures are Taiwan government reporting. They should not be treated as independent recorded future measurements. The wider lesson applies across the region. Internet facing infrastructure, health care, energy, and communications require particular attention during periods of geopolitical tension. The wider threat model also requires a wider response. First, combine intelligence across cyber, influence, physical security, and artificial intelligence. These teams may see different parts of the same operation. Second, map the links among domains, accounts, false identities, infrastructure, and narratives. One indicator may appear harmless. A connected pattern may reveal coordinated activity. Private organizations should not conduct offensive disruption on their own. Disruption may require support from technology platforms, industry partners, law enforcement, or government priorities. Third, study long term effects. Do not measure only whether one poster one campaign became popular. Strategic influence can work slowly through repetition. Fourth, build trusted communication channels before a crisis. Staff, partners, customers, and government contacts should know where reliable information will come from. Finally, connect every defensive action to a clear intelligence requirement. Ask what the actor wants, ask why your organization may matter, and ask which access information or relationship would help that objective. That is the central message of this briefing. Chinese threat activity is broad, but it is not random. Strategic context helps us identify risk earlier and respond with greater precision. I want to thank everyone, who attended this webinar today. We are out of time, but if you have any further questions, please follow-up with the team. I hope this was useful, and you can find additional context about recorded futures research on China on our website for free. So, press the. It will start from the bottom. Cool. Okay. So, the first question is, based on the the Chinese industrial policies, what will be the, level of threat, that the Japanese machine manufacturing and robotics, companies are facing in terms of their CAD data and control firmware? Incredibly high. So yeah. So the answer is incredibly high. So manufacturing companies operating in this sector are of strategic importance to China because the intellectual property is explicitly mentioned as one of the primary strategic objectives for this five year plan. The evidence backs this up. We identify consistent and persistent Chinese threat activity targeting Japanese manufacturing companies. While this threat activity is not unique to Japan, we do see it in other jurisdictions. Japan still maintains a competitive advantage in that intellectual property that China seeks for its strategic objectives. We assess the threats against Japanese manufacturing companies, particularly those in the robotics and related industries to be high over the next five years. Thank you. So I'm I'm gonna read it in Japanese first. So the question is, as they attack to, Taiwan, becoming persistent, what changes do you see in the trend of attacks, targeting Japanese semiconductor equipment and, material manufacturers, especially with APT effects? So we still can see a high, level and volume of APT attacks targeting Japanese semiconductors. Well, the volume is not as high as Chinese targeting of Taiwan or The United States. Japan still faces a moderate risk of APT targeting towards its semiconductor industry. Japan's strategic economic relationship with The United States and Taiwan is framed internally within China as an adversarial relationship to Chinese interests. Therefore, Taiwan, The United States, and Japan all sit within the same periphery of Chinese APT targeting. We expect the volume of APT attacks on Japanese semiconductor Thank you. So the question is, how should we, control the trade off, between, data sharing for r and d, time reduction and, the reduction of speed of development, resulting from, introducing zero trust? It's all about third party risk. As we know, r and d and manufacturing in general is a supply chain. R and d is dependent on trade relationships, international economic agreements, university agreements and research partnerships, transportation, logistics, and so many other international An organization operating in the r and d space may do everything right to defend its own network. But that does not mean that your partners are not also vulnerable. As we pivot away from internal zero trust and onto external threat intelligence, that third party supply chain risk is going to become far more important than your own internal security posture. This requires extensive intelligence beyond typical supplier and partner questionnaires. We are talking about external intelligence of infrastructure compromise, database breaches, ransomware and cyberattacks, and even APT scanning on your third party vendors. This is the direction that we must go in order to secure the Japanese r and d industry. Thank you. Sure. So next question is, with regards to, SharePoint attacks, is, activities, related to, targeting SharePoint online and Microsoft three sixty five tenants increasing in addition to on premise environment? Yes. They are. So those third party and cloud providers and those service providers continue to face disproportionate targeting by Chinese APT groups. These services are often easier to target due to compromised credentials, lack of multifactor authentication, device personal device malware infections, and other factors that make them easier targets for espionage actors. We continue to see disproportionate targeting and increased targeting of cloud services from 2024 into 2025, and we expect to see even more into 2026 and beyond. Thank you. So, complicated question, but we already do see Chinese APT targeting of the Japanese financial sector. We're not looking at wide scale theft or financial motivation. We're often looking at information about contracts, partners, suppliers, and international business dealings that may prove to be intelligence requirements for Chinese security services. We very rarely see Chinese actors, especially those that are state linked, involved in direct revenue generation unlike, say, North Korean actors targeting Japan. The targeting of Japanese finance and banking is far more strategic than pure financial motivations. We're looking at procurement networks, contracts, who pays who, who is a customer of that bank, and other important information that serves Chinese intelligence requirements and not purely revenue generation. Thank you. So based on the, current, threat environment, what will be the three most important initiatives that the Japanese and Asia manufacturers, should implement as priority in the coming twelve months? And, also, what will be the first target for APD tax? Absolutely. So answering the second question first, yet again, your third party suppliers are likely going to be the biggest targets. That is going to be then your first defensive priority, assessing that third party risk and those supply chain risks. The second priority is going to be securing personal identities, so your employer identities, credentials, devices, particularly in your access to cloud and third party services. If that's many identity The third priority will be assessing your access and your organization's exposure to the open Internet and legitimate Internet services. Those are the three biggest ways that Japanese organizations remain vulnerable to Chinese APT groups, third party compromise, identity exposure, and then exposure to legitimate Internet services. Prioritizing those three areas will defend you against the overwhelming majority of Chinese APT attacks. Thank you.