{
  "id": 7481450,
  "title": "AI kill switch: Who gets to pull the plug if AI becomes too dangerous?",
  "url": "https://urgent.news/2026/09/15/ai-kill-switch-who-gets-to-pull-the-plug-if-ai-becomes-too-dangerous",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-15T04:53:28.000Z",
  "source": {
    "name": "The Indian Express",
    "slug": "the-indian-express",
    "url": "https://indianexpress.com/article/technology/artificial-intelligence/ai-kill-switch-who-gets-to-pull-the-plug-10878375/"
  },
  "original_language": "en",
  "account": "The recent developments in the AI industry have sparked a heated debate over the need for a kill switch to regulate the rapid advancement of frontier artificial intelligence. CEO of Anthropic, Dario Amodei, has called for a deliberate slowing down of AI development and the establishment of independent third-party evaluators to monitor the technology. However, the question remains: who gets to pull the plug if AI becomes too dangerous? Jack Clark, co-founder of Anthropic, believes that an AI kill switch may need to be mandatory for companies developing frontier AI technology. While various AI labs have their own methods to shut down AI systems, Clark argues that lawmakers must make such safeguards mandatory. The specifics of the kill switch, including technical requirements and enforcement, should be part of the broader policy debate surrounding AI. Former AI researcher Jacob Coxon has suggested that top AI labs are racing towards self-improving superintelligence, which could potentially eliminate humanity by the end of the decade. In response, Amodei has urged for the pace of AI development to be slowed down and closely monitored. Nonetheless, security researcher Gary McGraw warns against the alarmist tone surrounding AI risks, emphasizing that the issue lies more with the engineering of the systems than their potential dangers. He argues that transparency, such as disclosing underlying architecture, training data, and testing regimens, is more crucial than dictating what companies can or cannot build. Nevertheless, the debate over AI development continues to polarize, with some advocating for slower progress and stronger oversight, while others dismiss the existential threat as overstated or merely a ploy for hype.",
  "summary": null,
  "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."
}