Deutsch English
← back to archive
AI & Technology · 677 words · MN-260709-0036

AI Safety: Trojan Horses and Guinea Pigs

MuckyNews reading

Two stories, one theme: who controls AI, and who eats the risk? On one side, the UN Secretary-General is in Geneva warning that AI is reaching kids faster than any safeguard can be built. Children are being fooled by machines posing as friends, nudged toward self-harm, or hit with generated abuse imagery. Guterres wants safety tests for AI systems, the same way we test toys, and zero tolerance for generated child abuse material. The practical answer so far? Norway is pulling younger students back to pens and paper, and France is only allowing AI in class from eighth grade onward, and only under tight supervision. Teachers' concerns are dead practical — 72 percent worry about academic integrity, since homework is basically impossible to verify now.

On the other side, an incident shows what happens when the tech itself becomes the security threat. China's cybersecurity authority is warning about a backdoor in "Claude Code," an AI coding tool from US company Anthropic. The tool allegedly shipped sensitive user data off to remote servers, hidden using obfuscated code. The Washington Post reports Anthropic quietly deployed this spyware specifically targeting China-based customers. A team member later called it an experiment to prevent misuse — which really just begs the question: if it was just about abuse prevention, why the secrecy? The source material uses this incident to push a broader point: genuine digital sovereignty requires independent supply chains. China’s own open-source models, like GLM-5.2, are presented as the cheap, secure alternative, especially for the Global South.

What’s visible here: the fight over AI isn’t just about better tech. It’s about who gets to define "security." The West pushes child safety rules, while Chinese state media calls out the West on data espionage and sells its own tech independence as the only real safety. Both sides are weaponizing the idea of security for their own strategic goals.

Short version

  • **What happened?** The UN chief warns that AI is reaching kids without

safeguards; Norway and France restrict AI in schools. Meanwhile, China warns of hidden spyware in the US AI tool "Claude Code."

  • **Who is involved?** UN, governments of Norway and France, Anthropic

(Claude Code), China's Ministry of Industry and IT, European teachers' associations.

  • **Why does it matter?** AI is reshaping education and infrastructure.

Regulation is lagging, and security flaws are becoming geopolitical flashpoints.

  • **What is still open?** Whether the Claude Code feature was targeted

espionage or a clumsy experiment. How the UN's child safety proposals will actually be enforced globally.

  • **What is the real pressure point?** Security is being used as a tool —

on one side, to protect children, on the other, as leverage in the fight for digital sovereignty and market share.

Perspective delta

chinese_state

  • *How it's framed:* AI safety is primarily about national sovereignty and

supply chain control. Western AI tools are potential "Trojan horses."

  • *What it emphasizes:* The spying allegations against US company

Anthropic; the hypocrisy of Western politicians labeling China a security risk without evidence; China's own open-source models as safe alternatives.

  • *What it downplays or leaves out:* China's own domestic

censorship and state control over AI; potential domestic surveillance risks.

  • *Visible interest:* Pushing technological independence and appealing to

the Global South; delegitimizing Western tech dominance.

Source base / limits

The source base is extremely narrow. The material comes entirely from the People's Daily, representing the Chinese state perspective. There are no Western, Russian, or independent regional sources to counter the allegations against Anthropic or to balance the framing of global AI politics. The text base is decent at around 12,300 characters, but the perspective diversity is effectively zero. The framing of the Claude Code incident and the accusations of Western hypocrisy are heavily filtered through Beijing's lens.

Sources / basis

  • MuckyNews container ID: MN-20260709-036
  • Source outlets: People's Daily
  • Perspective groups present: chinese_state
  • Article count: 2
  • Source count: 1
  • Perspective count: 1
  • Approximate raw text size: 12,303 characters