Anthropic Reveals Global Misuse of Claude AI — China at the Center of Military and Intelligence Operations

Reuters reports that multiple actors, including state‑linked groups, attempted to bypass Claude’s safeguards to obtain sensitive military information, run simulations, and automate cyber‑intrusion workflows. The cases highlight how modern AI has become a force multiplier for authoritarian regimes and conflict zones worldwide.

China: AI‑Driven Weapons Research and Targeting Simulations

Several of the most concerning cases involve actors tied to Chinese research institutions and the People’s Liberation Army (PLA).

Targeting Taiwan’s defenses

One actor used Claude to develop electronic warfare software that prioritized 12 military targets in Taiwan, including early‑warning radars, Patriot missile batteries, and air bases. Anthropic traced the activity to the PLA Academy of Military Sciences, suggesting a direct link to China’s military modernization efforts.

Anti‑torpedo systems

In another case, Claude generated a 200‑page technical specification for an anti‑torpedo system for the Chinese navy. The output included comparisons with U.S. technology and simulated expert critiques — effectively providing a blueprint‑level analysis.

Reverse engineering foreign weapons

A Chinese actor also used Claude to analyze components of foreign high‑power microwave weapons and map out supply chains, apparently to support reverse‑engineering projects and briefings for senior CCP and PLA officials.

These incidents fit a broader pattern: the CCP increasingly leverages AI not only to control its population but also to accelerate military innovation and expand its strategic reach across the Indo‑Pacific.

Global Misuse: Russia, Iran, North Korea, Mali, Yemen

Anthropic’s report outlines additional cases that show how widely AI is being weaponized:

• Russia: Developers used Claude to program autonomous “kamikaze drones” capable of selecting targets without human oversight.

• Russian intelligence: A group resembling APT29 used Claude to automate nearly an entire cyber‑espionage workflow — from phishing to data exfiltration against Ukrainian and European targets.

• Iran: State‑linked actors used Claude for covert influence operations and psychological campaigns, including content aligned with IRGC interests.

• North Korea: Claude was used for cybercrime and cryptocurrency theft.

• Mali: A consultant developed a surveillance system capable of monitoring 25 million SIM cards, including call and text content.

• Yemen: Claude assisted in software development for rockets, ballistic systems, and hypersonic glide vehicles. After a failed test, the actors asked Claude to diagnose the problem.

Together, these cases show how AI is becoming embedded in global conflict dynamics — especially in environments where authoritarian control or instability already exists.

Anthropic: AI Is Not a “Game Changer,” but a Dangerous Accelerator

Anthropic emphasizes that AI models like Claude do not create entirely new capabilities. Instead, they accelerate existing operations, making them faster, more scalable, and more accessible.

Threat actors used multi‑step prompts, obfuscated goals, and parallel sessions to circumvent safeguards. The danger lies not in superintelligence, but in the automation of expertise that previously required trained specialists.

The CCP Context: Technology as a Tool of Control and Expansion

The report’s China‑linked cases align with a long‑documented pattern:

• The CCP uses AI for mass surveillance, censorship, and political control.

• AI‑driven repression has targeted groups such as Falun Dafa practitioners, Uyghurs, and other minorities.

• China’s military‑tech ecosystem blends civilian research, state security, and PLA modernization — creating a high‑risk environment for global stability.

In this context, the misuse of Claude is not an isolated incident but part of a broader strategy of digital authoritarianism.

Why These Findings Matter for the Free World

1. AI‑enabled weapons development lowers barriers for states and militias.

2. Cyber operations become more scalable and harder to detect.

3. Automated propaganda threatens democratic discourse worldwide.

4. Authoritarian regimes gain new tools to expand power internally and externally.

Democracies must respond by:

• establishing international AI safety standards,

• building shared threat‑intelligence databases,

• enforcing rigorous testing and auditing of high‑risk models,

• and openly identifying and sanctioning state actors who misuse AI.

Conclusion

Anthropic’s report underscores how real and immediate the risks of modern AI have become. China stands at the center of a global pattern of AI‑driven weapons development, espionage, and digital influence. For democratic societies and human‑rights movements — including Falun Dafa — the findings highlight the urgent need for transparency, accountability, and responsible technological governance.

.