{
  "id": 15819940,
  "title": "Anthropic pitches AI safety bill that may block a model if 50+ people could be killed",
  "url": "https://urgent.news/2026/10/20/anthropic-pitches-ai-safety-bill-that-may-block-a-model-if-50-people",
  "topic": "ai",
  "section": "AI",
  "published": "2026-10-20T23:01:23.000Z",
  "source": {
    "name": "TechCrunch",
    "slug": "techcrunch",
    "url": "https://techcrunch.com/2026/10/20/anthropic-pitches-ai-safety-bill-that-may-block-a-model-if-50-people-could-be-killed/"
  },
  "original_language": "en",
  "account": "Over the past year, worries have grown about the potential hazards associated with artificial intelligence, leading to frequent discussions about the appropriate course for AI safety. Recently, Anthropic has reportedly unveiled a proposal for a new bill aimed at AI safety, which it hopes may be adopted by Congress. The bill outlines specific circumstances under which an AI model's distribution could be prohibited. These scenarios include a \"foreseeable, imminent risk\" where more than 50 individuals could be potentially killed, or if property damage exceeding a billion dollars were to occur. The risk must stem from the creation or release of a biological weapon, the model evading its developer's controls, or a cyberattack, according to Politico. The proposed metrics, 50 people and a billion dollars, are comparable to those in a recent California law, Senate Bill 53, which also classified such situations as \"catastrophic risk.\" The bill suggests that independent evaluators, registered with the U.S. Commerce Department, could assess a model's potential risk of catastrophe. These evaluators may be compensated by the AI companies, as long as the payment cannot be contingent upon the evaluation results. Upon determining an imminent catastrophic risk, evaluators would have 72 hours to notify the Commerce Department. The Department of Commerce would then need to request a D.C. judge to halt the model's release. Importantly, the legislation would prevent any federal agency from independently halting the distribution of the model. Anthropic has not yet commented on the draft bill, but a spokesperson confirmed its existence to Politico. The proposal, part of Anthropic's Advanced AI Framework, has been advocated for with federal and state lawmakers since June. The bill's concerns are somewhat abstract compared to the current focus of the tech industry, which has been primarily centered on the risk of malicious hacking. However, companies like Anthropic have offered their own solutions to address the crisis they themselves have helped create. Despite warnings that their technology could lead to human extinction, labs have generally resisted mandatory government oversight, instead proposing voluntary systems of embedded evaluators to monitor models for potential risks or abuse. Anthropic's bill, however, appears to be a step further than other industry proposals, including Nvidia CEO Jensen Huang's assertion that the AI industry does not require any new laws.",
  "summary": "The proposed policy seems to offer protections against AI run amok but actually forbids federal agencies from blocking a model's release without the permission of a judge.",
  "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."
}