{
  "id": 7505624,
  "title": "AI safety requires more than just slowing our pace | Stuart Russell",
  "url": "https://urgent.news/2026/09/15/ai-safety-requires-more-than-just-slowing-our-pace-stuart-russell-7505624",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-15T08:00:31.000Z",
  "source": {
    "name": "Guardian Technology",
    "slug": "guardian-technology",
    "url": "https://www.theguardian.com/commentisfree/2026/sep/15/ai-safety-requirements"
  },
  "original_language": "en",
  "account": "Safety measures in the realm of artificial intelligence are non-negotiable, focusing on meeting concrete goals rather than merely adjusting the timeline of advancement. A recent week in AI has seen considerable drama due to the resignation of AI safety researcher Jacob Coxon from Anthropic. This followed a series of unsettling revelations regarding the OpenAI/Hugging Face incident. One Business Insider message, with the subject \"AI doomsday debate reaches boiling point,\" reflects the urgency of the situation.\n\nIn response to these concerns, AI safety researcher Stuart Russell received a lengthy letter from Anthropic CEO Dario Amodei, titled \"We Must Pace the Frontier.\" This 3,800-word letter aims to outline proposals for avoiding or postponing potential doomsday scenarios. Sam Altman of OpenAI, Elon Musk of xAI, Demis Hassabis of Google DeepMind, and Satya Nadella of Microsoft have all expressed support for these ideas.\n\nThe concept of \"pacing the frontier\" appears to be borrowed from the world of Formula 1 racing, where a pace car sets a slow pace on the track for safety reasons. Amodei explains that pacing does not imply halting model training or technical progress. Instead, his proposal consists of three key points: having third-party evaluators work inside each company with full access to the systems; establishing common safety standards and limits on the rate of unchecked AI progress in democratic countries; and creating a broader compact to involve authoritarian countries as well.\n\nWhile Amodei’s proposal seems to suggest a slowdown in the rate of progress, he clarifies that it doesn't mean stopping model training or technical advancements. Instead, the approach is to set safety requirements first and then only proceed with further development once those requirements are met. This is similar to the \"red lines\" approach advocated by AI safety researchers, where developers must demonstrate compliance with safety standards before moving forward.",
  "summary": "Safety requirements are non-negotiable. They depend on meeting concrete goals, not just adjusting a timeline It has been a week of high drama in AI, precipitated by the resignation of the AI safety researcher Jacob Coxon from Anthropic. This followed several weeks of increasingly lurid and disturbing revelations about the OpenAI/Hugging Face incident. My inbox yesterday included a message from…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "The Guardian US",
        "title": "AI safety requires more than just slowing our pace | Stuart Russell",
        "url": "https://urgent.news/2026/09/15/ai-safety-requires-more-than-just-slowing-our-pace-stuart-russell",
        "published": "2026-09-15T08:00:31.000Z"
      }
    ]
  },
  "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."
}