Anthropic Reveals Massive Chinese Anti-Dissident AI-Generated Slander
Anthropic released a threat‑intelligence report on September 11 detailing how its AI model Claude was weaponized for cyberattacks, influence operations, surveillance, biological misuse, and weapons development. The report identifies systematic misuse of Claude by China‑based actors linked to the CCP’s security, propaganda, and religious‑affairs apparatus. Targeted groups include dissidents, journalists, religious minorities, and diaspora communities worldwide, with specific mention of journalists for “Bitter Winter”. The findings highlight the global reach of AI‑driven threats and the involvement of state and non‑state actors in weaponizing advanced AI for political and ideological purposes.
15 days ago
Anthropic’s September 11 threat-intelligence report describes attempts by China-based actors to use Claude for surveillance, transnational repression, propaganda, cyber operations, military development, and targeting of dissidents and minority communities. The operations focused particularly on religious groups, Uyghur activists and journalists, overseas democracy advocates, and diaspora communities; Anthropic says it disrupted all of the activity identified in the report.
Anthropic said the report was its most detailed assessment to date of efforts by state and non-state actors to misuse Claude for cyberattacks, influence operations, surveillance, biological misuse, and weapons development. It also warned that AI allows individuals using stolen API keys or public multi-agent tools to perform tasks that previously required teams of specialized operators.
The article emphasizes China-based activity aligned with the Chinese Communist Party’s security, propaganda, and religious-affairs priorities. The reported targets included Catholic leaders, Tibetan Buddhists, Falun Gong practitioners, Taiwanese Christians, Uyghurs, journalists, dissidents, and overseas activist communities.
One operation allegedly used Claude to generate Chinese-language dossiers, personnel-research drafts, investigative reports, and daily situational briefings on religious figures and communities. The activity included collecting birth and immigration information, social-media accounts, and details about religious venues, including floor plans and structural diagrams.
The operators reportedly instructed Claude to adopt official Chinese government terminology, including describing Tibet’s administration in exile as an illegal separatist administration and Falun Gong as an “evil cult.” Their monitoring covered Chinese platforms such as WeChat, Xiaohongshu, Douyin, and Weibo, as well as international services including LinkedIn, Instagram, Threads, X, and Facebook.
Another operation involved public-security and state-security personnel using Claude Code and customized tools to monitor sentiment, query a government surveillance database, and prepare daily reports on politically sensitive incidents. The article says one police academy student used the system to identify ten private citizens for possible “control,” including petition prevention, coercive summonses, and monitoring of movements and communications.
According to the report as described in the article, a local Chinese state-security bureau used Claude to prepare briefings on overseas activists and gather intelligence about potential protest venues. The reported examples included pro-democracy demonstrations in Vancouver, Uyghur cultural events in Türkiye, and Oslo Freedom Forum screenings.
Automated reporting pipelines allegedly scraped civil-society outlets and classified Uyghur advocacy as connected to terrorism while labeling major human-rights organizations hostile forces. The article connects this terminology to the Chinese Communist Party’s stated approach of psychological, legal, and public-opinion warfare.
A separate China-based contractor reportedly used Claude to produce government-style “public opinion monitoring” briefings. The system processed 15 to more than 30 articles per day from platforms including Weibo, X, YouTube, Telegram, and Facebook, scored their political sensitivity, and rewrote criticism of China using mandated terminology. Some resulting documents allegedly recommended enforcement measures available only to government authorities.
Anthropic also identified an operation in Syria that allegedly tracked, profiled, and attempted to recruit Uyghurs and Uyghur armed formations. The actor sought people who might have access to Uyghur units within the Syrian Army, offering payment for intelligence, while also mapping Uyghur businesses and locations in Syria. Anthropic reportedly assessed that the actor was probably a contractor working for Chinese state security rather than a state organ acting directly.
The article describes a separate campaign against Uyghur journalists, including journalists who write for Bitter Winter. The operators reportedly organized data from WhatsApp groups and Telegram channels to identify people vulnerable because of financial hardship, family separation, or ideological disillusionment. They also sought targets with relatives in Xinjiang, planned coordinated mass-reporting campaigns, and intended to use foreign-front delegitimization and bot networks against diaspora journalists.
The report allegedly identified a China-based actor using Claude to prepare a fire-control specification and acquisition documents for an undersea anti-torpedo weapons system. The actor produced a 200-page technical proposal, an executive presentation, and comparisons with U.S. Navy systems; Anthropic assessed that the actor was connected to China’s defense industry.
The same broader set of activities reportedly included drafting surveillance-platform tenders and capability brochures for bureau-level Chinese government customers, indicating a possible contractor-to-government procurement model. The article says the actor repeatedly tried to bypass Claude’s safeguards after the system refused requests involving covert interrogation scripts and mass fake personas.
Anthropic said it disrupted all of the operations described in the report and used the findings to improve its detection systems and safeguards. The article also notes that Anthropic identified parallel activity by Russian and Iranian actors, but portrays the China-related operations as particularly broad, bureaucratic, and closely connected to Chinese Communist Party security priorities.
The reported operations demonstrate the use of a foreign frontier AI model to automate intelligence collection, political monitoring, influence campaigns, recruitment efforts, and technical work. The article presents the disruption of the campaigns as a positive outcome while describing their existence as evidence that China’s domestic repression mechanisms are being extended into overseas and digital environments.
Assess Catholic doctrine on state‑controlled propaganda and dissent
Catholic doctrine rejects state-controlled propaganda when it monopolizes information, falsifies reality, suppresses legitimate debate, or manipulates citizens. It does not, however, condemn every public-information campaign or every effort by government to communicate. The decisive tests are truth, human dignity, the common good, and freedom of responsible judgment.
Catholic teaching distinguishes legitimate persuasion from propaganda in the morally corrupt sense. A campaign may be justified when it serves truth, uses methods compatible with human dignity, and advances the public interest.
Propaganda becomes unacceptable when it:
Thus, the problem is not simply that the state communicates an official position. The problem arises when political authority treats citizens as objects to be managed rather than persons capable of seeking truth. The Catechism particularly condemns totalitarian systems that systematically falsify truth, control opinion through the media, repress opposing thought, and manufacture public trials.
A state therefore may inform citizens, argue for public policies, and encourage the common good, but it may not claim exclusive control over truth or eliminate independent sources of information. Pius XII criticized the silencing of Catholic publications as though truth were subject to the exclusive discretion of political rulers.
Catholic social teaching supports broad freedom to express and compare serious opinions, provided that the common good and public morality are not endangered. Open discussion is necessary for society to assess competing claims and reach sound practical judgments.
This does not mean that every opinion is equally true or morally permissible. Religious freedom, for example, is grounded in the dignity of the person and the need to seek truth without coercion; it is not based on the claim that all religions or teachings have equal value. Nor is freedom a moral license to embrace error.
Accordingly, Catholic doctrine rejects both:
The state must respect conscience and may not arbitrarily compel people to act against it. Civil authority exists for the common good, and its coercive power is legitimate only when ordered to a genuinely just purpose.
But conscience is not merely personal preference. It is not what creates right and wrong; it is an instrument for recognizing moral truth and interpreting norms that precede the individual’s choice. Therefore, appeals to “conscience” do not automatically justify disobedience, nor does political dissent become morally right simply because it is sincerely held.
When a government commands something contrary to fundamental moral truth, a person must not cooperate in the evil. The prudential question of how to resist—public criticism, conscientious objection, lawful protest, or civil disobedience—depends on the circumstances and the likely effects on the common good. The supplied material supports the primacy of truth and conscience, but does not establish a universal formula for when civil disobedience is obligatory.
This must be distinguished from dissent from Catholic teaching. Catholics may question, discuss, or respectfully investigate prudential applications and disciplinary decisions. But a claimed “right to dissent” cannot mean selectively accepting or rejecting the Church’s moral teaching according to personal preference. John Paul II specifically warned that conscience does not establish moral truth but recognizes and applies it.
The Catholic position can therefore be summarized as follows:
In short, Catholic doctrine opposes both propaganda that enslaves the mind and dissent that treats personal autonomy as the source of moral truth. Truth must guide both political authority and individual conscience.