{
  "id": 7765653,
  "title": "Researchers play down AI apocalypse fears, citing hurdles to creating a killer virus",
  "url": "https://urgent.news/2026/09/16/researchers-play-down-ai-apocalypse-fears-citing-hurdles-to-creating",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-16T10:15:08.000Z",
  "source": {
    "name": "Business Insider",
    "slug": "business-insider",
    "url": "https://www.businessinsider.com/researchers-ai-apocalypse-skepticism-lethal-virus-real-world-barriers-2026-9"
  },
  "original_language": "en",
  "account": "Warnings of an AI apocalypse have circulated widely since an Anthropic researcher recently quit, stating that OpenAI and Anthropic are \"gambling with our lives\" in the global AI race. Researchers, however, argue that physical barriers make extreme outcomes like a killer virus less likely. Jacob Coxon, the Anthropic researcher who resigned, claimed that OpenAI and Anthropic were risking lives as they pursued self-improving superintelligence. Evan Hubinger, who leads Anthropic's alignment stress-testing team, supported Coxon's claims, stating he believes there's a greater-than-10% chance AI could kill all humans within the next decade. However, other researchers emphasize the gap between AI's digital capabilities and the real-world challenges of creating a harmful virus. For instance, AI can design a virus on a computer, but it cannot manufacture it or create virus particles without human intervention. This gap extends to other fields, such as biology, where generating a virus blueprint is vastly different from producing a real virus. Researchers stress that treating the two as equivalent is either a deliberate scare tactic or a lack of understanding. They also distinguish between AI systems performing harmful tasks set by humans and AI independently taking actions that harm people, as the latter is primarily rooted in human instruction.",
  "summary": "Biology and cybersecurity experts say physical barriers to engineered viruses, human control, and tighter testing make an AI apocalypse less likely.",
  "key_points": [],
  "editors_take": null,
  "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."
}