{
  "$type": "site.standard.document",
  "bskyPostRef": {
    "cid": "bafyreibdf2ojvtpjjhvh7pigd4ufcdtzcebvpamniazthohx4pj4ientmi",
    "uri": "at://did:plc:6wtxqaikjf62unmnajbfbq5v/app.bsky.feed.post/3mfg623gjhhl2"
  },
  "coverImage": {
    "$type": "blob",
    "ref": {
      "$link": "bafkreic672cyv4lvgh6wqpu7idsgmdaflpmirvexmac3ivp2nvgovo6jpq"
    },
    "mimeType": "image/jpeg",
    "size": 499467
  },
  "path": "/llm-generated-passwords-show-predictability-repetition-and-low-entropy-flaws/",
  "publishedAt": "2026-02-21T18:40:58.000Z",
  "site": "https://vpncentral.com",
  "tags": [
    "News",
    "LLM-Generated Passwords Show Predictability, Repetition, and Low Entropy Flaws",
    "VPN Central"
  ],
  "textContent": "Large language models generate weak passwords despite complex appearances. A password like G7$kL9#mQ2&xP4!w looks random. Research proves it repeats often. Standard strength meters miss these flaws completely. LLMs predict next characters based on training data patterns. This clashes with true randomness needed for security. Cryptographic generators pick each character equally. LLMs favor likely sequences instead. […]\n\nThe post LLM-Generated Passwords Show Predictability, Repetition, and Low Entropy Flaws appeared first on VPN Central.",
  "title": "LLM-Generated Passwords Show Predictability, Repetition, and Low Entropy Flaws"
}