{
  "id": 6615359,
  "title": "Anthropic Staffers Again Sound the Alarm on AI Catastrophe",
  "url": "https://urgent.news/2026/09/10/anthropic-staffers-again-sound-the-alarm-on-ai-catastrophe",
  "topic": "ai",
  "section": "AI",
  "published": "2026-09-10T17:43:42.000Z",
  "source": {
    "name": "Mother Jones",
    "slug": "mother-jones",
    "url": "https://www.motherjones.com/politics/2026/09/anthropic-ai-risks-coxon-amodei-openai-cybersecurity-danger/"
  },
  "original_language": "en",
  "account": "On Tuesday, former AI researcher Jacob Coxon announced his resignation from the AI firm Anthropic, warning that both Anthropic and OpenAI were \"gambling with our lives\" by developing models that could rapidly improve themselves until they reach \"superintelligence.\" In a viral social media post, Evan Hubinger, who leads a team responsible for testing Anthropic's safety, acknowledged that his colleagues \"really do earnestly believe AI could kill all humans!\" This isn't the first time researchers have raised concerns about AI's dangers, but Coxon's post gained significant attention. While the intricacies of advanced models can be difficult to grasp, several key facts about AI highlight the reasons for alarm. Firstly, AI is already surpassing human abilities in important domains. Secondly, leading companies are rapidly improving these models. Lastly, no one knows how to reliably keep AI behavior aligned with human goals. Recently, OpenAI shared a solution to one of the six remaining \"Millennium Problems,\" a significant mathematical accomplishment reliant on advanced models. This, combined with AI's growing capabilities in critical areas like cybersecurity and novel virus design, raises concerns about the potential risks of more generalized artificial intelligence. Companies and governments are racing to develop better AI models, but there is no clear timeline for achieving \"recursive self-improvement,\" where top models can rapidly build better versions of themselves surpassing human intelligence. At the same time, safety researchers have long been worried about the challenge of aligning AI behavior with human values, especially as AI systems become more capable.",
  "summary": "On Tuesday, experienced AI researcher Jacob Coxon resigned from the AI firm Anthropic—saying that both that company and OpenAI, his previous employer, were “gambling with our lives” by developing models that could improve themselves at a rapid clip until they reach “superintelligence.” In an alarming social media post, Evan Hubinger, who leads a team that […]",
  "key_points": [
    "Former AI researcher Jacob Coxon resigned from Anthropic, warning of potential AI catastrophe.",
    "AI surpasses humans in critical domains, but alignment with human goals remains unknown."
  ],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 2,
    "also_reported_by": [
      {
        "outlet": "Mother Jones",
        "title": "Anthropic Staffers Again Sound the Alarm on AI Catastrophe",
        "url": "https://urgent.news/2026/09/10/anthropic-staffers-again-sound-the-alarm-on-ai-catastrophe-6757642",
        "published": "2026-09-10T17:43:42.000Z"
      }
    ]
  },
  "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."
}