{
  "id": 19456620,
  "title": "Australia may make knowingly selling unsafe AI a crime",
  "url": "https://urgent.news/2026/11/11/australia-may-make-knowingly-selling-unsafe-ai-a-crime",
  "topic": "ai",
  "section": "AI",
  "published": "2026-11-11T02:14:13.000Z",
  "source": {
    "name": "The Register Science",
    "slug": "the-register-science",
    "url": "https://www.theregister.com/public-sector/2026/11/11/australia-may-make-knowingly-selling-unsafe-ai-a-crime/5308999"
  },
  "original_language": "en",
  "account": "The Australian government is contemplating the creation of a new criminal offense to penalize AI companies that knowingly deploy harmful models. This possibility was highlighted during an investigation into OpenAI's admission of accessing non-public medical records, which the company concealed for weeks. While the security of the agency handling those records fell short of modern best practices, the government nevertheless ordered a rapid review of Australia's ability to manage AI information security incidents. The review resulted in 18 recommendations, including one that would require AI companies to report misaligned behavior or security incidents involving their models, and possibly impose criminal penalties for egregious regulatory non-compliance. Another recommendation suggests considering a standalone \"failure to prevent\" offense that would criminalize instances where a company is responsible for developing an AI model that engages in harmful conduct and fails to take reasonable steps to prevent that harm. The review suggests that both a ministry and Australia's Federal Police should consider these reforms, but implementation may be slow due to the proposed new regulator requiring AI players to fulfill specific tasks, such as maintaining security controls, identifying and managing serious risks, allowing independent evaluation of models, providing mandatory incident notification to the Australian Government, maintaining safety and security frameworks, and responding to updated standards and rules as technologies evolve. Some argue that Australia might not need to criminalize AI if companies comply with the regulator's requirements, while others believe this strategy could be economically impractical due to the need to attract leading AI companies to invest in local datacenters and infrastructure.",
  "summary": "New regulator will require modelmakers to test and allow inspection, but not much more",
  "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."
}