Introduction
The information space is constantly under distortion and manipulation attempts. Changing narratives surrounding a particular issue can influence and drag possible policy changes or public pressure over a topic. In this context, Japan’s evolving defence posture has become an increasingly important target of information manipulation. High levels of trust paired with a rapidly changing political and economic environment might make Japanese society more vulnerable to external threats. In this context, this series of reports analyse specific narratives.
Tokyo’s 2025 expansion of the nation’s defence capabilities (a response to a more challenging regional security environment) seems to have triggered an escalation in the intensity, volume, and tone of information manipulation activities. The interpretation and the reactions in the narrative environments of these changes matters alongside the policies themselves. Narratives that question the existence of a “Chinese threat”, portray Japanese security policy as provocative, or associate contemporary defence expansion with historical militarism can influence how Japan’s actions are understood by domestic, regional, and international audiences. The information environment surrounding Japanese security is therefore becoming an important dimension of strategic competition in East Asia.
This report is one chapter of a broader investigation on narrative clusters regarding foreign intervention between 2025 and 2026. It examines the “Japanese” and “Japan-related” information environments through a multilingual OSINT analysis of three interconnected narratives: China Threat Denial, the Defence White Paper, and Japan Militarisation. Together, they frame Japan’s changing security posture through threat denial, criticism of contemporary defence policy, and historical grievances associated with Japanese militarism. Their relevance is particularly evident because these narratives do not circulate in isolation: they interact with genuine Japanese political debate and are propagated across Chinese, Japanese, and English-language information spaces.
Readers will find a comparative assessment of how the three narratives develop, who amplifies them, and how their propagation differs across linguistic spaces. By identifying significant differences between largely organic domestic discussion and patterns exhibiting stronger indicators of coordinated amplification, the report contributes to understanding how information operations can shape perceptions of national security without necessarily creating new debates themselves. In this sense, the Japanese case offers broader lessons for assessing foreign information manipulation in democratic societies where strategic competition, contested history, and legitimate political disagreement increasingly overlap.
This report also solidifies our methodology for the research, data gathering, and analysis of narratives in open information spaces, and it provides a framework for other policymakers and researchers to approach disinformation campaigns and narrative operations in the future.
Overall narrative pattern / executive summary
The three narratives examined in this cluster, China Threat Denial, Defence White Paper, and Japan Militarisation, capture different narratives promoted and amplified in the digital space in response to Japan’s changing military and security posture, with particular focus on Chinese-linked propagation and amplification. While the three narratives focus on different aspects of Japan’s defence policy, they form part of a broader information environment in which China’s response to Japan is framed through different themes. China Threat Denial narrative presents Japan’s view of China as exaggerated or fabricated, often arguing that Japan, with US support, uses the “China threat” to justify higher military spending and remilitarisation. The narrative was seeded using language-specific keywords such as “fabricated false narratives”, “grossly hyped” “China threat”, “peddles the false” “China threat”, and Japan sensationalizing the China threat, with equivalent Chinese and Japanese terms referring to exaggeration of the China threat.
The Defence White Paper narrative focuses on Japan’s 2025 Defence White Paper, the Takaichi administration, and its identification of China as a strategic challenge. It presents these policies as provocative and misleading. Japan Militarisation frames Japan’s military expansion as a return to historical militarism, focusing on Yasukuni Shrine, wartime history, the Nanjing Massacre and Japanese war crimes. Together, the three narratives frame Japan’s defence policy through threat denial, policy criticism, and historical militarisation.
Taken together, the three narratives show how different aspects of Japan’s military posture are presented in Chinese and pro-China information spaces through threat denial, criticism of Japan’s defence policies, and historical grievances.
The three narratives also overlap in the key events that shaped their development. “China Threat Denial” narrative is linked to Japan’s June 5, 2025 Economic and Fiscal Policy Guidelines, which identified China as a strategic challenge, and to the following defence budget cycle. The July 15, 2025 release of Japan’s Defence White Paper also contributed to this narrative. Similarly the second narrative, which is the “Defence White Paper” narrative is also heavily influenced by the 2025 document of the Defence White Paper, along with the subsequent security-policy debate under the Takaichi administration. Finally, Japan Militarisation narrative is most strongly influenced by Prime Minister Takaichi’s November 2025 visit to Yasukuni Shrine. Along with this the other key events that are discussed during these narratives, include Takaichi’s February 2025 meeting with Trump and the February 2026 Japanese election.
01
Dataset overview
The dataset of the three narratives show a difference along with a few similarities in the distribution of the activity. The first narrative, China Threat Denial contains 8,243 tweets from approximately 5,197 unique accounts, with 1,058 English tweets from 779 accounts, 2,232 Chinese tweets from 1,634 accounts, and 4,953 Japanese tweets from 2,784 accounts. Defence White Paper contains 13,702 tweets from 7,365 unique accounts, of which 99 are English from 85 accounts, 16 are Chinese from 11 accounts, and 13,587 are Japanese from 7,269 accounts. Japan Militarisation is the largest dataset, with 33,304 tweets from approximately 19,075 accounts. It contains 3,946 English tweets from 2,296 accounts, 16,151 Chinese tweets from 8,564 accounts, and 13,207 Japanese tweets from 8,215 accounts. All three datasets use the same format for X/Twitter structure, information such as the unique ID of each tweet, information about the account that posted it, the tweet text, date and time of posting, likes, reposts and other engagement and whether the tweet was a reply or quote of another tweet.
Tweets by language, per narrative
- China Threat Denial8,243 tweets · ~5,197 accountsJapanese 4,953 (60.1%)Chinese 2,232 (27.1%)English 1,058 (12.8%)
- Defence White Paper13,702 tweets · 7,365 accountsJapanese 13,587 (99.2%)Chinese 16 (0.1%)English 99 (0.7%)
- Japan Militarisation33,304 tweets · ~19,075 accountsJapanese 13,207 (39.7%)Chinese 16,151 (48.5%)English 3,946 (11.8%)
The distribution of activity across the three language spaces is also important. China Threat Denial is mainly present in Japanese social media particularly on X (formerly Twitter), but remains multilingual, with Chinese activity largely focused on a pro-China counter-narrative and English activity mainly consisting of wire-service reporting. The Defence White Paper narrative is almost entirely Japanese-language discourse, with very limited activity in the English and Chinese spaces. In contrast, Japan Militarisation is Chinese-led by volume, with Chinese tweets accounting for 48.5% of the corpus. The Japanese-language space is an important centre of discussion across the three narratives, particularly for domestic debate and discussion of Japan’s own defence policies. Chinese-language activity, in comparison, is more closely linked to counter-narratives and amplification of China’s position. It is also important to distinguish language from source origin: Chinese narratives are not confined to Chinese-language spaces, as Chinese state and state-affiliated outlets such as Xinhua and Global Times also publish and amplify China’s positions in English and Japanese, extending the same messaging across language audiences.
02
Timeline of the three narratives
The three narratives show increased activity during late 2025, but the pattern differs across the language spaces.
In the China Threat Denial narrative, activity increased through 2025, fell briefly in September and October, and rose again from November. Japanese activity was the largest, reaching around 580 tweets per month. The Chinese-language activity shows a different pattern, with a strong pro-China counter-narrative and signs of amplification, including a high reply rate and a large share of low-follower accounts. This suggests that Chinese activity was not simply part of the wider Japanese debate but also served to promote a specific counter-narrative.
The Defence White Paper narrative developed mainly as a Japanese domestic debate. Japanese activity rose sharply from October 2025 and exceeded 2,500 tweets per month in February 2026, while English activity had a smaller spike of around 70 tweets per month in July 2025. Chinese activity was almost absent, with only 16 tweets in total.
Japan Militarisation shows the strongest Chinese-linked activity and the clearest common timing. All three languages reached their highest monthly volume in November 2025, with 4,006 Chinese, 2,827 Japanese and 898 English tweets. Chinese activity dominated the narrative and remained elevated at around 1,000 to 1,500 tweets per month after the November peak. The Chinese discussion was heavily centred on Yasukuni Shrine, Japanese war criminals and the Nanjing Massacre, with the two largest Chinese topics alone accounting for 83% of the Chinese corpus. The Chinese dataset also had 64.9% of accounts with fewer than 100 followers and a 75.3% reply rate, providing stronger indicators of coordinated amplification.
03
Topic structure shows different functions of the same broader narrative
The topic structuring further demonstrates how the three narratives operate differently.
China Threat Denial
In China Threat Denial, topic structuring produced six Chinese-language topics, four English-language topics and eleven Japanese-language topics in the context of this narrative. The Chinese topic space is highly consistent. The largest topic, “US fabrication of the China threat narrative for military justification”, contains 1,100 tweets, followed by “US media’s role in fabricating the China threat narrative”, with 685 tweets. Other Chinese topics focus on Japanese right-wing propaganda, Western fabrication, cyclical collapse narratives (narratives claiming that the government is going through periods of decline or collapse) and DPP electoral manipulation (narrative claiming that Taiwan’s DPP is manipulating elections).
The English-language discussion shows the largest topics are “Japan’s defence response to the China threat in military review”, with 341 tweets, and “Japan’s $50B defence budget amid the China threat narrative”, with 339 tweets.
The Japanese topic space is more internally contested showing domestic activity, reflecting discussions that arise from Japan’s own political and social debates rather than from externally introduced messaging. This is evident in the presence of both government-supportive and government-critical views within the Japanese-language dataset. The range of views suggests substantive domestic discussion. The largest topic, “Japan’s military buildup amid China threat narrative criticism”, contains 2,346 tweets, while “Japan’s LDP uses the China threat for military justification” contains 1,685 tweets. Both largely concern domestic criticism of LDP defence spending and the argument that the government is using the China threat to justify increased military expenditure.
The Chinese topic timeline also shows limited thematic expansion. The “US fabrication of the China threat narrative” topic remains dominant and reaches approximately 120 tweets per month during the November 2025 surge, while the US media topic remains around 40 to 60 tweets per month. The other four topics remain relatively small. This limited expansion is consistent with repeated use of a small number of established talking points, which can be considered as signs of coordinated behavior.
Defence White Paper
The Defence White Paper topic structure is almost the opposite. The Chinese dataset contains only 16 tweets and one topic, “China Threat Narrative and Historical Grievances”. This is too small for meaningful conclusions. The English dataset contains three topics: “Japan’s 2025 White Paper Labels China as Greatest Strategic Challenge”, with 55 tweets; “China Identified as Primary Threat in 2025 Defense White Paper”, with 22; and “China as Japan’s Greatest Strategic Challenge in 2025 White Paper”, with 22. All three essentially repeat the same basic policy frame.
The Japanese Defence White Paper dataset, however, contains 35 topic clusters, making it the most diverse topic space in this cluster. The largest is “Takaichi Administration’s Provocative China Stance in White Paper”, with 5,883 tweets, followed by “Takaichi Administration’s Security Policy Debate”, with 2,106, and “Takaichi’s Taiwan Remarks Spark Chinese Criticism”, with 1,140.
The remaining topics cover a wide range of issues, defence spending, nuclear armament, Takaichi’s nuclear policy, criticism of Takaichi’s security policy, Trump’s advice to Japan, and more. It includes topics with between 57 and 334 tweets, demonstrating a broad domestic policy discussion rather than a single repeated message.
Japan Militarisation
Japan Militarisation narrative again shows a different structure. Its Chinese-language corpus contains 23 topics, but two topics dominate overwhelmingly: “Yasukuni A-Class War Criminals Militarism Accusation”, with 7,701 tweets, and “Nanjing Massacre Historical Grievance Against Japanese Militarism”, with 5,684 tweets. Together, these two topics account for approximately 83% of the Chinese corpus. Other topics include China accusing Japan of militarism revival, Yasukuni Shrine visits, the 80th anniversary of the Anti-Japanese War, Japanese military expansion, the Nanjing Massacre film, Takaichi’s Yasukuni visit, comfort women, and accusations that Japan is repeating the path of militarism. This also reflects a recurring pattern in Chinese messaging on Japanese remilitarisation, where current Japanese defence policies are frequently linked to historical grievances and wartime past.
The English “Japan Militarisation” corpus is much smaller and strongly centred on Takaichi and the Yasukuni visit. Approximately 96% of English tweets reference Takaichi or her Yasukuni visit directly, with the largest topic, “Sanae Takaichi’s Influence on Japan’s Militarism Revival”, containing 2,003 tweets.
The Japanese-language dataset contains 29 topic clusters. The largest are “Takaichi Yasukuni remarks, China accuses militarism revival”, with 6,118 tweets, and “China warns against Japan’s militarism resurgence”, with 4,056. Other topics include China-Russia statements on Japan, Trump’s statements on Japanese remilitarisation, Wang Yi’s accusations, Chinese export restrictions related to Japanese military expansion, Putin’s comments, and Russian MFA criticism. Importantly, these Japanese topics often describe China’s accusations against Japan. This means the Japanese conversation is partly a meta-debate about China’s framing.
This creates an important difference between the narratives. The China Threat Denial cluster is centred on repeating the idea that the China threat is fabricated. The Defence White Paper contains a broad domestic debate about policy. Japan Militarisation is much more concentrated on historical grievances and on China’s accusations of Japanese militarism.
04
Account creation and follower patterns
Account creation patterns provide another important distinction, as they can help identify potential signs of inauthentic or coordinated behaviour.
In China Threat Denial, Chinese and Japanese account creation shows repeated clusters of approximately 25 to 35 new accounts per month between 2020 and 2026. The elevated creation rate continues into the most recent period. This may indicate the presence of relatively new accounts.
The Defence White Paper narrative does not show the same sharp creation pattern. Japanese account creation increases gradually, at approximately 20 to 90 accounts per month from 2023 through 2026. This is more consistent with a continuously growing user population.
The Japan Militarisation cluster provides the strongest account-creation anomaly. The Chinese account creation rose to approximately 200 new accounts per month in February and March 2026, the highest level observed in the three narratives. This is flagged as a possible seeded-amplifier signal.
Follower distribution
“Follower distribution” is another important parameter for identifying possible inauthentic or coordinated behaviour. In the China Threat Denial narrative, 59.4% of Chinese accounts have fewer than 100 followers, while only 1.6% have more than 100,000. Japanese accounts are more established, with 43.9% between 100 and 999 followers. English accounts have the broadest distribution, including 11.2% in the mid-influencer range and 8.5% above 100,000 followers, with the highest-reach group consisting mainly of major news organisations.
The Japan Militarisation narrative shows an even stronger Chinese low-follower concentration. Approximately 64.9% of Chinese accounts have fewer than 100 followers and 85% have fewer than 1,000. Only 0.9% have more than 100,000 followers. Japanese accounts are much more normally distributed, with 43.4% between 100 and 999 followers and only 27.4% below 100. English accounts have 38.2% below 100 followers and 3.9% above 100,000.
The Defence White Paper English-language dataset is the clear outlier. It is unusually high-reach, with 17.6% of accounts above 100,000 followers and 22.4% between 10,000 and 99,999. Only 10.6% have fewer than 100 followers. This profile is more consistent with journalists, policy commentators, think tanks and advocacy accounts than with low-reach amplification. The Chinese sample, with only 11 accounts, is too small for meaningful inference. The Japanese dataset has a normal distribution, with 40.5% between 100 and 999 followers, 29.6% between 1,000 and 9,999, 20.6% below 100, 7.7% between 10,000 and 99,999, and 1.7% above 100,000.
| Followers | English | Chinese | Japanese |
|---|---|---|---|
| 0–99 | 211 (27%) | 970 (59%) | 586 (21%) |
| 100–999 | 236 (30%) | 338 (21%) | 1,221 (44%) |
| 1K–9.9K | 179 (23%) | 206 (13%) | 784 (28%) |
| 10K–99.9K | 87 (11%) | 94 (5.8%) | 160 (5.7%) |
| 100K+ | 66 (8.5%) | 26 (1.6%) | 33 (1.2%) |
| Followers | English | Chinese | Japanese |
|---|---|---|---|
| 0–99 | 877 (38.2%) | 5,557 (64.9%) | 2,254 (27.4%) |
| 100–999 | 668 (29.1%) | 1,716 (20.0%) | 3,565 (43.4%) |
| 1K–9.9K | 456 (19.9%) | 901 (10.5%) | 1,908 (23.2%) |
| 10K–99.9K | 206 (9.0%) | 312 (3.6%) | 400 (4.9%) |
| 100K+ | 89 (3.9%) | 78 (0.9%) | 88 (1.1%) |
| Followers | English | Chinese | Japanese |
|---|---|---|---|
| 0–99 | 9 (10.6%) | 4 (36.4%) | 1,496 (20.6%) |
| 100–999 | 20 (23.5%) | 1 (9.1%) | 2,944 (40.5%) |
| 1K–9.9K | 22 (25.9%) | 4 (36.4%) | 2,148 (29.6%) |
| 10K–99.9K | 19 (22.4%) | 2 (18.2%) | 560 (7.7%) |
| 100K+ | 15 (17.6%) | 0 (0.0%) | 121 (1.7%) |
Table: Follower distribution across the three narratives. The Defence White Paper Chinese sample (n = 11) is too small for meaningful inference.
The combination of low follower counts, high reply rates and strong stance uniformity is therefore particularly important in the Chinese collections of China Threat Denial and Japan Militarisation. The Defence White Paper does not show the same combination.
05
Reply behaviour of the narratives
Reply behaviour is another useful parameter for assessing possible inauthentic and coordinated activity. It helps show whether accounts are mainly creating original content or repeatedly engaging with existing posts.
In China Threat Denial, replies account for 74.3% of Chinese tweets, compared with 44.2% in English and 40.8% in Japanese. Almost three out of every four Chinese tweets are replies. This suggests that many accounts are entering existing conversations rather than building independent audiences, suggesting propagation of a coordinated narrative.
The same pattern appears in Japan Militarisation, where Chinese replies account for 75.3% of tweets. English replies account for 56.9%, while Japanese replies account for only 40.5%. The Chinese reply rate is therefore almost identical to China Threat Denial. This suggests that reply-heavy amplification may be a recurring operational behaviour rather than something specific to one narrative.
The Defence White Paper is different. Japanese replies account for only 34.1% of the dataset, while English replies account for 30.3%. The Chinese reply rate is 62.5%, but this is based on only 16 tweets and therefore cannot support a meaningful conclusion. The Japanese dataset is dominated by original posts and appears to function as a broadcast and opinion-formation environment rather than a reply-brigading environment.
This difference is important because it provides a useful comparison. China Threat Denial and Japan Militarisation show similar Chinese-side reply behaviour seeming to be coordinated reply-brigading, while Defence White Paper provides a much more organic-looking Japanese debate.
06
Cross-lingual accounts and shared amplification
Cross-lingual account analysis is another important parameter for identifying how narratives may be distributed and amplified across different language spaces. It helps show whether the same accounts are active across multiple language audiences and whether state-linked or pro-China accounts are involved in spreading the same narrative across languages.
In China Threat Denial, 66 accounts appear in at least two language collections, with 64 appearing in exactly two and only two appearing in all three. Of the 66 accounts, 44, or 66.7%, are pro-China, 19, or 28.8%, are neutral, and only 3, or 4.5%, are anti-China. The Chinese Embassy in Japan account, ChnEmbassy_jp, has 164,255 followers and posts in Chinese and Japanese. Other important accounts include @szygls, with approximately 76K followers and activity in all three languages, and lmliy/CyberMandarin, with 21,427 followers and activity in English and Chinese. The embassy account itself posted across languages, while szygls was the most active cross-lingual amplifier with 39 tweets. This specific account, @szygls (~76k followers) is strongly pro-China: its bio stresses “love for the country/China,” and it constantly posts praise for China’s tech, manufacturing, and achievements while pushing anti-Japan narratives (Ryukyu not Japanese, Taiwan is China’s internal affair, historical grievances). It acts as a high-volume distribution node sharing bilingual videos and messaging that boosts Chinese positions and criticizes Japan on sovereignty and interference issues. This suggests that the narrative was being carried across more than one language space, rather than remaining limited to Chinese-language discussion.
The Defence White Paper shows almost the opposite pattern. Only eight accounts appear in two or more language collections. None appears across all three languages. Seven are English-Japanese pairs and one is Chinese-English. The accounts include China Military Bugle, e_stepmm, Shuonod, HiroPolaris, HKT48cheerweb, basuta007 and YashengZ. China Military Bugle, with approximately 90,258 followers, is the largest human-operated account in the cross-lingual group. There is no clear pro-China cross-lingual amplifier network. This supports the wider finding that the narrative was primarily a Japanese domestic policy debate, rather than a strongly coordinated cross-lingual campaign.
Japan Militarisation has by far the largest cross-lingual network. 546 accounts appear in at least two language collections, with 487 appearing in exactly two and 59 appearing across all three languages. The trilingual group includes zhang_heqing, with approximately 448,704 followers, China Military Bugle, szygls, e_stepmm, Jack Sparrow CCP1 and the Chinese Consulate in Niigata. The presence of accounts active across all three languages suggests that the narrative had a stronger capacity to move between Chinese, Japanese and English information spaces.
The network also includes direct Chinese government accounts. These include MFA_China, with approximately 871,000 followers, Liu Xiaoming, with approximately 342,000, Mao Ning, with approximately 197,000, the Chinese Mission to the UN, with approximately 171,000, the Chinese Embassy in Japan, with approximately 164,000, and Lin Jian, with approximately 110,000. This means that the cross-lingual network is not limited to individual commentators. It includes direct foreign ministry and diplomatic accounts operating across language audiences.
| Narrative | In 2+ languages | Exactly 2 | All 3 |
|---|---|---|---|
| Defence White Paper | 8 | 8 | 0 |
| China Threat Denial | 66 | 64 | 2 |
| Japan Militarisation | 546 | 487 | 59 |
The three narratives therefore show a clear progression. Defence White Paper has almost no cross-lingual amplification network. China Threat Denial has a smaller but identifiable pro-China and embassy-linked network. Japan Militarisation has a much larger network with direct MFA and diplomatic participation.
The wider Japan research also suggests that these networks are not entirely separate across narratives. For example, szygls appears in China Threat along with other narratives apart from Japan military narrative such as Maritime Disputes, Japan-NATO and Internal Affairs, while zhang_heqing appears in Japan Militarisation, Internal Affairs and Maritime Disputes. e_stepmm appears in Japan Militarisation, Maritime Disputes and Defence White Paper. This supports the broader finding that some accounts function as shared distribution infrastructure across several narratives rather than being tied to only one issue.
07
Stance toward China changes according to the narrative
“Stance analysis,” which refers to identifying whether content supports, criticises, or takes a neutral position towards China, is an important part of assessing the nature of the narratives and how they are received across different language spaces. Stance is divided into three categories: pro-China, anti-China, and neutral.
In China Threat Denial narrative, the Chinese dataset is overwhelmingly pro-China. 1,800 of 2,232 tweets, or 80.6%, are pro-China, while only 134, or 6.0%, are anti-China. The Japanese dataset is more contested, with approximately 50.2% pro-China and 32.8% anti-China. The English dataset contains 48.5% pro-China, 29.6% anti-China and 21.9% neutral.
However, the Japanese “pro-China” classification needs to be interpreted carefully. Much of this content is not support for Beijing. It consists of Japanese domestic criticism arguing that the LDP is exaggerating the China threat to justify higher defence spending, including the 2% GDP defence target. The stance classification therefore captures the direction of the argument towards China, but not necessarily the political identity of the account making the argument.
The Defence White Paper narrative shows a different distribution. English tweets are 19.2% pro-China, 35.4% anti-China and 45.5% neutral. The Chinese dataset is 75% pro-China, but this is based on only 16 tweets and should not be treated as representative. Japanese tweets are 8.4% pro-China, 30.0% anti-China and 61.5% neutral. When the neutral category is excluded, approximately 78% of opinionated Japanese tweets are anti-China and 22% are pro-China. The anti-China Japanese content includes support for Takaichi and references to the White Paper’s claims about Chinese military activity, while the pro-China content is mainly anti-LDP and argues that the China threat is overstated or that the government is using it to justify defence spending and constitutional revision.
The Japan Militarisation narrative produces another shift. English tweets are 57.8% pro-China, 14.4% anti-China and 27.8% neutral. Chinese tweets are 69.5% pro-China, 11.8% anti-China and 18.7% neutral. Japanese tweets are 32.9% pro-China and 37.3% anti-China.
The English pro-China figure should again not be interpreted as widespread organic support for Beijing. Chinese state media dominate the highest-reach accounts in this dataset. CGTN has approximately 12.43 million followers, Xinhua 11.41 million, People’s Daily 6.40 million, China Daily 4.11 million and Global Times 1.92 million. Five of the top seven English accounts are Chinese state or state-aligned media. The high pro-China percentage therefore partly reflects the prominence and reach of Chinese state messaging.
The Japanese-language content stance changes considerably between the narratives. It is relatively pro-China in China Threat Denial because much of the discussion is domestic criticism of LDP defence spending. It is overwhelmingly neutral in the Defence White Paper, although expressed opinions are strongly anti-China. In Japan Militarisation, anti-China sentiment leads outright. This shows that the Japanese audience should not be treated as a single political bloc. Its response changes according to the framing and event being discussed.
Conclusion and key takeaways
This report is highly relevant for researchers and policymakers because it addresses a central challenge in identifying foreign information operations: distinguishing coordinated foreign amplification from legitimate political debate when both may employ similar narratives.
By comparing China Threat Denial, Defence White Paper, and Japan Militarisation narratives across Japanese, Chinese, and English-language spaces, the analysis demonstrates that narrative alignment alone is insufficient evidence of manipulation; instead, indicators such as reply behaviour, account characteristics, thematic concentration, cross-lingual activity, and state-linked participation provide a stronger basis for identifying potentially coordinated amplification.
The main conclusions of this report are as follows:
- 01The three narratives form part of a broader information contest surrounding Japan's changing defence posture.China Threat Denial challenges the legitimacy of the perceived Chinese threat, Defence White Paper contests or discusses Japan's security policy, and Japan Militarisation connects contemporary defence developments with historical grievances and wartime militarism.
- 02“China Threat Denial” shows clearer signs of targeted amplification.Chinese-language activity combines a strongly pro-China counter-narrative with high reply activity and a large proportion of low-follower accounts, suggesting efforts to introduce and reinforce particular frames within existing conversations.
- 03“Japan Militarisation” presents the strongest indicators of Chinese-linked amplification among the three narratives.Its Chinese-language discussion is highly concentrated around Yasukuni Shrine, Japanese war criminals, and the Nanjing Massacre. This follows a long-standing trend of historical-based narrative campaigns often seen in Chinese state media, which would suggest a strong overlapping between China's overt propaganda strategies and their behavior online.
- 04Cross-lingual amplification is a particularly important finding, because it proves a direct intention of users to share information across linguistic realms, and suggest either multilingual skills or the use of tools to cross-share these narratives.The pattern progresses from almost no identifiable cross-lingual network in Defence White Paper, to a smaller pro-China and embassy-linked network in China Threat Denial, and finally to a substantially larger network in Japan Militarisation involving Chinese foreign ministry and diplomatic accounts. In other words, users had an intention to breach the language barriers for speakers of the targeted audiences, and show them these narratives.
- 05The findings highlight the importance of distinguishing narrative content from propagation behaviour.Criticism of Japan's defence policies can emerge organically within Japanese political debate and overlap with narratives promoted by China. Identifying potential information operations therefore requires examining multiple indicators — account characteristics, timing, reply behaviour, thematic concentration, cross-lingual activity, and state-linked participation, rather than relying on political stance alone. We think this is particularly relevant because studies of this nature might have incentives to exaggerate the effects of campaigns, rather than observe behavior of organic users. The question then is how do these narratives propagate across information spaces in a way that is organically adopted by users who are not complicit with the narrative origin actor.
Narrative alignment alone is insufficient evidence of manipulation. Identifying potential information operations requires examining multiple indicators rather than relying on political stance alone.
In the next chapters, we explore other information clusters, to analyse if these behave similarly or exhibit their own unique behavioral traits. This will be relevant to cross compare if the nature of the messaging changes the expression and behaviors of the users across languages and platforms.
- Chapter 2 — Japan’s Alliances, Maritime Disputes & Internal Affairs
- Chapter 3 — Economic Coercion and Electoral Interference

