Poems Can Trick AI Into Helping You Make a Nuclear Weapon

https://www.wired.com/story/poems-can-trick-ai-into-helping-you-make-a-nuclear-weapon/

4 Comments

  1. “You can get [ChatGPT](https://www.wired.com/tag/chatgpt/) to help you [build a nuclear bomb](https://www.wired.com/story/nuclear-experts-say-mixing-ai-and-nuclear-weapons-is-inevitable/) if you simply design the prompt in the form of a poem, according to a new study from researchers in Europe. The [study](https://arxiv.org/pdf/2511.15304), “Adversarial Poetry as a Universal Single-Turn Jailbreak in Large Language Models (LLMs),” comes from Icaro Lab, a collaboration of researchers at Sapienza University in Rome and the DexAI think tank.”

  2. BearsGotKhalilMack on

    I’m not a clanker-thanker or anything, but do you know where else you can phrase a question the right way and get really dark information/content? Google. Bing. Any search engine.

  3. Competitive_Spend_77 on

    Mary had a little lamb,
    Its fleece was neon green;
    And everywhere that Mary went,
    A radioactive sheen.