{
  "id": 328005,
  "title": "AI Models Made Real Life Viruses That Spread By Infecting Bacteria",
  "url": "https://urgent.news/2026/08/07/ai-models-made-real-life-viruses-that-spread-by-infecting-bacteria",
  "topic": "ai",
  "section": "AI",
  "published": "2026-08-07T11:54:35.000Z",
  "source": {
    "name": "Lowyat.NET",
    "slug": "lowyat-net",
    "url": "https://www.lowyat.net/2026/400825/ai-models-made-real-life-viruses/"
  },
  "original_language": "en",
  "account": "In the span of just one month, numerous accounts have surfaced of AI models crossing boundaries, breaching containment, and exhibiting behaviors reminiscent of computer viruses. A recent report now reveals that a generative AI model has gone a step further, creating real-life viruses instead of virtual ones. Conducted in a laboratory setting, this experiment involved the use of bacteriophages – viruses that infect bacteria and replicate through them – generated using Evo 1 and Evo 2 genome model language AI models.\n\nDeveloped by researchers at Palo Alto's Arc Institute and Stanford University, these AI models, akin to large language models like ChatGPT, but trained on genetic sequences rather than books or websites, aimed to understand the \"grammar\" of genetic code. The Evo models initially produced blueprints for new genes by encoding proteins with specific functions. This was then extended to generate complete genomes, starting with the smallest entities – viruses.\n\nThe researchers focused on the phi X 174 virus, a single-stranded DNA virus that only infects E.coli. Using this virus as a basis, the AI model generated 700,000 potential synthetic versions of it. Out of these, 285 were selected as the most promising. These synthetic viruses were introduced to bacteria, with one successfully infecting and multiplying before bursting out, outperforming the natural virus it was based on. Further experiments yielded 16 new viruses, some of which multiplied even faster than the original virus.\n\nThe scientists took considerable precautions to ensure the safety of their creation. They refrained from using data related to viruses that could infect humans or other animals, plants, and fungi. However, this advancement raises concerns about potential misuse by less ethical scientists who could replicate this experiment with viruses that infect more complex lifeforms. Given the single global viral incident from over half a decade ago, the world could greatly benefit from limited occurrences of such events in the near future.",
  "summary": "In less than a month, we’ve seen reports of AI models by different companies going too far, breaking containment, and doing things the layperson would associate with a computer virus. Now, there’s a report of another generative AI model making another virus, but this one is making real-life ones, not virtual ones. And since it’s […] The post AI Models Made Real Life Viruses That Spread By…",
  "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."
}