{
  "id": 6671139,
  "title": "Artificial Intelligence: AI Company: Activities Surrounding Possible Biological Weapons Research",
  "url": "https://urgent.news/2026/09/11/kunstliche-intelligenz-ki-firma-aktivitaten-rund-um-mogliche",
  "topic": "business",
  "section": "Business",
  "published": "2026-09-11T00:07:56.000Z",
  "source": {
    "name": "Handelsblatt",
    "slug": "handelsblatt",
    "url": "https://www.handelsblatt.com/technik/it-internet/kuenstliche-intelligenz-ki-firma-aktivitaeten-rund-um-moegliche-biowaffen-forschung/100253729.html"
  },
  "original_language": "de",
  "account": "Anthropic, a KI firm, has blocked several attempts by researchers to use its software Claude for possible development of bioweapons. The blocked attempts came from scientists in undisclosed regions with ties to local government agencies. Anthropic reported five cases of attempted misuse of its KI models related to bioweapons research, with researchers seeking organisational support such as paper writing rather than scientific help. The firm noted that while KI models are designed to refuse assistance with biological weapon development, it can be challenging to distinguish between harmless and malicious intentions in research with both medical and military applications. The blocked attempts included efforts to research a virus, a variant of bird flu transmissible to mammals, and a toxin.",
  "summary": "Concerns about artificial intelligence include that it could be used to develop biological weapons. An AI company has found that scientists are seeking help with questionable research.",
  "key_points": [
    "Anthropic blocks researchers using Claude for biowarfare research.",
    "Attempts reveal covert access by scientists from blocked regions.",
    "AI models struggle to distinguish between harmless and harmful intentions."
  ],
  "editors_take": "Anthropic's actions against researchers probing its AI for biowarfare applications mark a stricter approach to policing dual-use research, highlighting challenges in distinguishing harmless from harmful intent in AI-assisted scientific endeavors.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}