[
  {
    "id": "attackers-will-use-machine-learning-2016",
    "date": "2016-12-13",
    "quote": "So the way this applies to infosec is like this: When attackers or defenders need to confirm a known thing, they might use Supervised Learning. And where they want to search for new ways to find attackers (or new victims) they will use Unsupervised Learning. … Expect both attackers and defenders to be using both with increasing frequency in the coming years.",
    "url": "https://danielmiessler.com/blog/how-attack-and-defense-can-leverage-supervised-and-unsupervised-learning",
    "source": "blog",
    "title": "How Cyber Attack and Defense Can Leverage Supervised and Unsupervised Learning",
    "theme": "consequences",
    "claim": "Attackers will increasingly use machine learning to qualify victims, triage fuzzing results and mine their scans for new targets.",
    "status": "came-true",
    "evidence": [
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/threat-actor-usage-of-ai-tools",
        "label": "Google GTIG documents threat actors using AI across the attack lifecycle, including the first malware that calls an LLM while it runs",
        "date": "2025-11-05"
      },
      {
        "url": "https://www.anthropic.com/news/disrupting-AI-espionage",
        "label": "Anthropic disrupts a state-sponsored campaign in which AI performed 80-90% of the work, including reconnaissance of target systems",
        "date": "2025-11-13"
      }
    ],
    "also_said": [],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "attack-automation-systems-crawl-the-internet-2018",
    "date": "2018-02-04",
    "quote": "So elite attack teams will basically rig up extraordinary automation systems that constantly crawl and parse the entire internet, with special focus on certain targets, and then take all the data they find and put it into a format that can be consumed by various types of algorithms. … So what we'll have is the battle of the algorithms.",
    "url": "https://danielmiessler.com/blog/future-iot-attack-defense-ai",
    "source": "blog",
    "title": "The Future of Attack and Defense is AI",
    "theme": "harnesses",
    "claim": "Small elite attack teams will build automation systems that crawl the whole internet and feed algorithms that pick targets, replacing thousands of human hunters.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/news/disrupting-AI-espionage",
        "label": "Anthropic: a state-sponsored group used Claude Code against about 30 targets, with AI doing 80-90% of the campaign and humans at only 4-6 decision points",
        "date": "2025-11-13"
      },
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold",
        "label": "Google GTIG: threat actors now run multi-agent offensive frameworks (ARTEX, CAIRN, PentAGI, CyberStrikeAI) in live intrusions",
        "date": "2026-11-16"
      }
    ],
    "also_said": [],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "attackers-win-the-llm-race-2023",
    "date": "2023-01-09",
    "quote": "Here's a fun question: who is going to be better at using ChatGPT and future models for cybersecurity attack and defense? Will it be the attackers or the defenders? My money is on the attackers for no reason other than them having higher numbers, more time, and more scrappiness. … Examples of AI-powered attacks will (and already are in some cases) include faster and better phishing campaigns, automated exploit code writing, automated reverse engineering, automated BEC and other social engineering, information warfare campaigns, etc.",
    "url": "https://danielmiessler.com/blog/unsupervised-learning-no-364-reality-headset-bingpt-aicyber",
    "source": "newsletter",
    "title": "Unsupervised Learning NO. 364",
    "theme": "consequences",
    "claim": "AI-powered attacks will include faster phishing and automated exploit writing, reverse engineering and BEC, and attackers will out-use defenders with these models.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/news/detecting-countering-misuse-aug-2025",
        "label": "Anthropic: a criminal used Claude Code to automate reconnaissance, credential harvesting and intrusion against at least 17 organizations, then extort them",
        "date": "2025-08-27"
      },
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/threat-actor-usage-of-ai-tools",
        "label": "Google GTIG documents threat actors using AI across the attack lifecycle, including the first malware that calls an LLM while it runs",
        "date": "2025-11-05"
      }
    ],
    "also_said": [
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-364-reality-headset-bingpt-aicyber",
        "title": "Unsupervised Learning Newsletter NO. 364",
        "date": "2023-01-09"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com/blog copy of the newsletter edition principal-authored; the quoted passage is Daniel's own commentary, not the news summary around it."
  },
  {
    "id": "ai-helps-attackers-first-2023",
    "date": "2023-07-21",
    "quote": "AI will most help the attacker side first, and then it will help defenders more in the long-term. … In the early days of AI, attackers will be able to use AI to automate attacks while defenders still lack context about their environment.",
    "url": "https://danielmiessler.com/blog/will-ai-help-moreattackers-defenders",
    "source": "blog",
    "title": "Who Will AI Help More—Attackers or Defenders?",
    "theme": "consequences",
    "claim": "AI will help attackers first: they will automate attacks for three to five years while defenders still lack the internal context to use AI well.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/news/detecting-countering-misuse-aug-2025",
        "label": "Anthropic: a criminal used Claude Code to automate reconnaissance, intrusion and extortion against at least 17 organizations",
        "date": "2025-08-27"
      },
      {
        "url": "https://www.ncsc.gov.uk/report/impact-ai-cyber-threat-now-2027",
        "label": "UK NCSC: AI-assisted vulnerability research and exploit development will expand intrusion capability through 2027",
        "date": "2025-05-07"
      }
    ],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/1696598994305036523",
        "title": "X post: Who gains the upper hand in cybersecurity – attackers…",
        "date": "2023-08-29"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "ai-agents-take-over-security-testing-2023",
    "date": "2023-11-07",
    "quote": "In the next 3-5 years we’ll see AI agents largely take over security testing.",
    "url": "https://x.com/DanielMiessler/status/1721931374255587594",
    "source": "tweet",
    "title": "X post, November 7, 2023",
    "theme": "harnesses",
    "claim": "Within three to five years, AI agents that execute and re-plan task lists will largely take over security testing.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://xbow.com/blog/top-1-how-xbow-did-it",
        "label": "XBOW's autonomous pentester reaches #1 on HackerOne's US leaderboard",
        "date": "2025-06-24"
      },
      {
        "url": "https://www.darpa.mil/news/2025/aixcc-results",
        "label": "DARPA AIxCC finals: autonomous systems find and patch vulnerabilities across 54 million lines of real-world code",
        "date": "2025-08-08"
      }
    ],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/1814211478670446866",
        "title": "X post: So if you want to know the future of…",
        "date": "2024-07-19"
      }
    ],
    "authorship": "own-account",
    "authorship_note": "Posted from Daniel's own X account (x.com/DanielMiessler); corpus attribution is \"unverified\" for all social posts. The quoted words are the post's own text, not a quote of anyone else."
  },
  {
    "id": "hacker-assistants-with-attack-frameworks-2023",
    "date": "2023-12-31",
    "quote": "Performs automated target discovery and service enumeration on all discovered targets … GLitCH has access to 31 of the best commercial attack and C2 frameworks. … can go all the way to launching live exploit code at targets (with permission, of course)",
    "url": "https://danielmiessler.com/blog/ai-predictable-path-7-components-2024",
    "source": "blog",
    "title": "AI's Predictable Path: 7 Components",
    "theme": "harnesses",
    "claim": "Dedicated hacker AI assistants will wrap attack and C2 frameworks, enumerate targets automatically and go all the way to launching live exploits.",
    "status": "came-true",
    "evidence": [
      {
        "url": "https://blog.checkpoint.com/executive-insights/hexstrike-ai-when-llms-meet-zero-day-exploitation/",
        "label": "Check Point: HexStrike-AI, an open framework driving LLM agents across 150+ offensive tools, is discussed for exploiting fresh Citrix bugs within hours",
        "date": "2025-09-02"
      },
      {
        "url": "https://www.straiker.ai/blog/cyberspike-villager-cobalt-strike-ai-native-successor",
        "label": "Straiker: Villager, an AI-native pentest framework billed as a Cobalt Strike successor, passes 10,000 PyPI downloads",
        "date": "2025-09-11"
      },
      {
        "url": "https://www.crowdstrike.com/en-us/blog/unknown-threat-actor-uses-artex-to-target-south-korean-finance/",
        "label": "CrowdStrike: an actor runs the AI-driven ARTEX offensive framework against South Korean financial institutions",
        "date": "2026-10-07"
      }
    ],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/1995944820838465967",
        "title": "X post: Like I said in 2023 and 2024, security will…",
        "date": "2025-12-02"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_github-aliasroboticscai-cybersecurity-activity-7401709767234379776-P1wg",
        "title": "LinkedIn post: Like I said in 2023 and 2024, Cybersecurity will…",
        "date": "2025-12-02"
      },
      {
        "url": "https://x.com/DanielMiessler/status/1995944822012817902",
        "title": "X post: This CAI project (hilarious since my DA's name is…",
        "date": "2025-12-02"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "ai-scales-moderate-skill-to-billions-of-targets-2024",
    "date": "2024-01-17",
    "quote": "To me this NSA story highlights that in both offensive and defensive security use cases, the main advantage of AI will not be its exceptional (superhuman) capabilities, but rather the ability to apply pretty-good-intern or moderate-SME level expertise to billions more analysis points than before.",
    "url": "https://newsletter.danielmiessler.com/p/ul-415",
    "source": "newsletter",
    "title": "Unsupervised Learning NO. 415",
    "theme": "consequences",
    "claim": "AI's main edge in offense and defense will be moderate expertise applied to billions more targets and data points, not superhuman skill.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/news/detecting-countering-misuse-aug-2025",
        "label": "Anthropic: a criminal used Claude Code to automate reconnaissance, credential harvesting and intrusion against at least 17 organizations, then extort them",
        "date": "2025-08-27"
      },
      {
        "url": "https://research.openanalysis.net/claude/codex/hacking/ai%20hacking/llm/redteam/policy%20violation/2026/06/16/compromised-claude-hacking.html",
        "label": "OALABS: a low-skill attacker using vague prompts had coding agents breach at least 14 companies",
        "date": "2026-06-16"
      }
    ],
    "also_said": [
      {
        "url": "https://danielmiessler.com/blog/ul-415",
        "title": "Unsupervised Learning NO. 415 (blog copy)",
        "date": "2024-01-17"
      }
    ],
    "authorship": "own-commentary-in-mixed",
    "authorship_note": "Newsletter editions are \"mixed\" in the corpus. This passage is Daniel's first-person 💡 comment on an NSA news item; the item summary above it is not quoted."
  },
  {
    "id": "agents-triage-millions-of-targets-2024",
    "date": "2024-01-29",
    "quote": "One of the most powerful uses of AI—and especially agents—will be going through millions of targets and figuring out what to attack, when, and using which technique. … Red will move much faster than blue on this",
    "url": "https://danielmiessler.com/blog/ul-417",
    "source": "newsletter",
    "title": "UL NO. 417: NSA’s Broker Buys, AI-Assisted Attacks, Companies Only Want Killers",
    "theme": "harnesses",
    "claim": "AI agents will sort through millions of targets, choosing what to attack, when and how, and attackers will adopt this far faster than defenders.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/news/disrupting-AI-espionage",
        "label": "Anthropic: AI ran reconnaissance and exploitation against about 30 targets, issuing thousands of requests, often several per second",
        "date": "2025-11-13"
      },
      {
        "url": "https://www.crowdstrike.com/en-us/blog/vapor-panda-uses-artex-to-scale-global-campaign/",
        "label": "CrowdStrike: VAPOR PANDA uses the AI-driven ARTEX framework to scale an espionage campaign across many organizations",
        "date": "2026-10-29"
      }
    ],
    "also_said": [
      {
        "url": "https://www.youtube.com/watch?v=JU-5mSeczzY",
        "title": "Killer Context: How AI Will Eat Security and Software - Daniel Miessler",
        "date": "2025-11-04"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2071209132804178317",
        "title": "X post: It’s now possible to send a single prompt into…",
        "date": "2026-06-28"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_this-is-what-ive-been-worried-about-the-activity-7522097165033336833-Z64a",
        "title": "LinkedIn post: This is what I've been worried about: the techniques…",
        "date": "2026-10-31"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/ul-417",
        "title": "UL NO. 417: NSA's Broker Buys, AI-Assisted Attacks, Companies Only Want Killers",
        "date": "2024-01-29"
      },
      {
        "url": "https://danielmiessler.com/blog/ul-432",
        "title": "Unsupervised Learning NO. 432 (blog copy)",
        "date": "2024-05-14"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/ul-432",
        "title": "Unsupervised Learning NO. 432",
        "date": "2024-05-14"
      },
      {
        "url": "https://danielmiessler.com/blog/post-ai-security-floor",
        "title": "The Post-AI Security Floor",
        "date": "2026-11-01"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com/blog copy of the newsletter edition principal-authored; the quoted passage is Daniel's own commentary, not the news summary around it."
  },
  {
    "id": "agent-teams-with-captured-tester-methodology-2024",
    "date": "2024-02-19",
    "quote": "The way forward on automated hacking is this: 1) teams of agents, 2) extremely detailed capture of human tester thought processes, lots of real-world examples, and time. I suspect that in 2-5 years, agent-based web hacking will be able to get 90% of the bugs we normally see submitted in web bug bounties.",
    "url": "https://newsletter.danielmiessler.com/p/ul-420",
    "source": "newsletter",
    "title": "Unsupervised Learning NO. 420",
    "theme": "harnesses",
    "claim": "Teams of agents loaded with captured tester methodology will find about 90% of typical web bug-bounty findings within two to five years.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://xbow.com/blog/top-1-how-xbow-did-it",
        "label": "XBOW: for the first time in bug bounty history, an autonomous penetration tester reaches the top spot on HackerOne's US leaderboard",
        "date": "2025-06-24"
      }
    ],
    "also_said": [
      {
        "url": "https://danielmiessler.com/blog/ul-420",
        "title": "Unsupervised Learning NO. 420 (blog copy)",
        "date": "2024-02-19"
      },
      {
        "url": "https://x.com/DanielMiessler/status/1900974584075350018",
        "title": "X post: In one sense bug bounty is going away because…",
        "date": "2025-03-15"
      }
    ],
    "authorship": "own-commentary-in-mixed",
    "authorship_note": "Newsletter editions are \"mixed\" in the corpus. This passage is Daniel's first-person 💡 comment under a paper summary about LLM agents hacking websites; the summary is not quoted."
  },
  {
    "id": "hacking-frameworks-shrink-the-exploit-window-2024",
    "date": "2024-07-01",
    "quote": "These frameworks will be used by all attackers and defenders, and the window between new vulnerabilities and either exploitation or mitigation will shorten dramatically",
    "url": "https://danielmiessler.com/blog/ul-439",
    "source": "newsletter",
    "title": "Unsupervised Learning NO. 439",
    "theme": "consequences",
    "claim": "AI hacking frameworks will spread across attackers and defenders, and the time from a new vulnerability to its exploitation will shrink dramatically.",
    "status": "came-true",
    "evidence": [
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/m-trends-2026/",
        "label": "Mandiant M-Trends 2026: mean time to exploit falls to an estimated -7 days, so exploitation routinely comes before the patch",
        "date": "2026-03-23"
      },
      {
        "url": "https://www.prnewswire.com/news-releases/cogent-research-exploits-outpace-scanner-detection-for-62-of-critical-vulnerabilities-as-ai-compresses-time-to-exploit-to-under-12-hours-302783104.html",
        "label": "Cogent Security, 69,000+ CVEs: average disclosure-to-exploit time collapsed from 125.3 days to 0.5 days in 16 months as AI compresses exploit development",
        "date": "2026-05-27"
      }
    ],
    "also_said": [
      {
        "url": "https://newsletter.danielmiessler.com/p/ul-439",
        "title": "UL NO. 439: Humans vs. AI in Prediction Markets",
        "date": "2024-07-01"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_ive-done-security-for-25-years-and-the-activity-7449901104756101120-b0H5",
        "title": "LinkedIn post: attackers are using AI to compress discovery-to-exploitation from weeks…",
        "date": "2026-04-14"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com/blog copy of the newsletter edition principal-authored; the sentence is the last line of Daniel's own 💡 comment under the Project Naptime item, which the blog copy sets in bold (the quote stops before the closing ** so it matches the body literally)."
  },
  {
    "id": "uncensored-local-models-mine-stolen-data-2024",
    "date": "2024-08-13",
    "quote": "But not with AI. Now you can put all those docs into a local Chroma database (vectorized), and ask it questions using an uncensored (and perhaps even fine-tuned) version of Llama 3.1. … So now, with some smart prompting, you can ask a set of 25 questions to such a dataset that pull out ALL your attack use-cases. And hell—even write the attack emails for you.",
    "url": "https://danielmiessler.com/blog/ul-445",
    "source": "newsletter",
    "title": "Unsupervised Learning NO. 445",
    "theme": "combination",
    "claim": "Attackers will load stolen files into uncensored local open models to find extortion material and write the attack emails, replacing rooms of human analysts.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.welivesecurity.com/en/ransomware/first-known-ai-powered-ransomware-uncovered-eset-research/",
        "label": "ESET: PromptLock ransomware runs the open-weight gpt-oss-20b locally through Ollama to inspect target files, exfiltrate and encrypt",
        "date": "2025-08-26"
      },
      {
        "url": "https://www.anthropic.com/news/detecting-countering-misuse-aug-2025",
        "label": "Anthropic: a criminal used Claude Code to automate reconnaissance, credential harvesting and intrusion against at least 17 organizations, then extort them",
        "date": "2025-08-27"
      }
    ],
    "also_said": [
      {
        "url": "https://newsletter.danielmiessler.com/p/ul-445",
        "title": "UL NO. 445: Vegas Dump, Legal Firm Hacks, AI Agent Ascension ",
        "date": "2024-08-13"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com/blog copy of the newsletter edition principal-authored; the quoted passage is Daniel's own commentary, not the news summary around it."
  },
  {
    "id": "open-source-ai-arms-our-enemies-2024",
    "date": "2024-11-04",
    "quote": "There’s an obvious tradeoff of risks here. On one hand, you have big model vendors hoarding all the IP and becoming EvilCorps, and on the other hand, we open source the intelligence that our enemies use to attack us.",
    "url": "https://danielmiessler.com/blog/ul-457",
    "source": "newsletter",
    "title": "UL NO. 457: China Builds a Military App Using Meta Llama",
    "theme": "open-models",
    "claim": "Open-sourcing capable models means handing adversaries the intelligence they will use to attack us.",
    "status": "came-true",
    "evidence": [
      {
        "url": "https://thehackernews.com/2025/07/cert-ua-discovers-lamehug-malware.html",
        "label": "CERT-UA: APT28 malware LAMEHUG uses the open-weight Qwen2.5-Coder-32B-Instruct model to generate its commands",
        "date": "2025-07-17"
      },
      {
        "url": "https://www.crowdstrike.com/en-us/blog/vapor-panda-uses-artex-to-scale-global-campaign/",
        "label": "CrowdStrike: China-nexus VAPOR PANDA runs its AI attack framework primarily on DeepSeek",
        "date": "2026-10-29"
      }
    ],
    "also_said": [],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com/blog copy of the newsletter edition principal-authored; the quoted passage is Daniel's own commentary, not the news summary around it."
  },
  {
    "id": "attackers-model-your-company-better-than-you-2025",
    "date": "2025-05-14",
    "quote": "So where this all takes us is that the top priority of attackers will be having better UEC models of your organization than you do. So it'll be a competition between your attacker and you, between who has the most accurate and up-to-date context for your company.",
    "url": "https://danielmiessler.com/blog/unified-entity-context",
    "source": "blog",
    "title": "Unified Entity Context: The AI Stack Everyone Is Building Without Realizing It",
    "theme": "consequences",
    "claim": "Attackers' top priority will be an AI-built model of your organization that is fresher and more accurate than your own.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/news/disrupting-AI-espionage",
        "label": "Anthropic: the AI agent inspected each target's systems and infrastructure and picked out the highest-value databases, far faster than a human team",
        "date": "2025-11-13"
      }
    ],
    "also_said": [
      {
        "url": "https://www.youtube.com/watch?v=UwTTcka1Wd8",
        "title": "The Future of Hacking is Context",
        "date": "2025-05-30"
      },
      {
        "url": "https://danielmiessler.com/blog/the-great-transition",
        "title": "The Great Transition",
        "date": "2026-02-28"
      },
      {
        "url": "https://x.com/DanielMiessler/status/1930024677948436490",
        "title": "X post: The future of hacking isn’t smarter tools—it’s better context.…",
        "date": "2025-06-03"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_the-future-of-hacking-is-context-activity-7335790279947354113-jA5i",
        "title": "LinkedIn post: The future of hacking isn’t smarter tools—it’s better context.…",
        "date": "2025-06-03"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/ul-483",
        "title": "Unsupervised Learning NO. 483",
        "date": "2025-06-03"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "autonomous-hacking-goes-common-2025",
    "date": "2025-06-23",
    "quote": "If you’re in security testing, or really anything in security, this should give you pause. My guess is that this level of autonomy is only possible privately right now, but that it’ll be super common in 6-18 months.",
    "url": "https://newsletter.danielmiessler.com/p/ul-486",
    "source": "newsletter",
    "title": "Unsupervised Learning NO. 486",
    "theme": "harnesses",
    "claim": "Fully autonomous hacking, then possible only in private systems, will be super common within 6-18 months.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://blog.checkpoint.com/executive-insights/hexstrike-ai-when-llms-meet-zero-day-exploitation/",
        "label": "Check Point: the open HexStrike-AI framework drives LLM agents across 150+ offensive tools and is discussed underground for exploiting fresh bugs within hours",
        "date": "2025-09-02"
      },
      {
        "url": "https://www.crowdstrike.com/en-us/blog/crowdstrike-uncovers-new-agentic-ai-campaigns/",
        "label": "CrowdStrike documents four previously unreported actors running agentic intrusion campaigns in which AI agents make decisions and execute commands",
        "date": "2026-10-21"
      },
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold",
        "label": "Google GTIG: agentic threat activity crosses a significant threshold as actors deploy autonomous multi-agent frameworks in live intrusions",
        "date": "2026-11-16"
      }
    ],
    "also_said": [
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-510",
        "title": "Unsupervised Learning NO. 510",
        "date": "2025-12-18"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2041975193233772728",
        "title": "X post: Any model that is this good at chaining together…",
        "date": "2026-04-08"
      }
    ],
    "authorship": "own-commentary-in-mixed",
    "authorship_note": "Newsletter editions are \"mixed\" in the corpus. This passage is Daniel's first-person comment under a news item about a fully automated bug-hunting company; it gives his own guess and timeline, and quotes no one."
  },
  {
    "id": "open-models-plus-shared-attack-scaffolding-2025",
    "date": "2025-11-19",
    "quote": "Perhaps the scariest thing to me is that the open-source models or the proprietary models used by China only have to get a certain level of quality before the scaffolding itself—which can be shared as a series of skills or as a complete automation system on GitHub or something—can be distributed to all attackers within their ecosystem and used in attacks at scale.",
    "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-507",
    "source": "newsletter",
    "title": "Unsupervised Learning NO. 507",
    "theme": "combination",
    "claim": "Once open models are good enough, attack scaffolding shared as skills or full automation systems will spread to every attacker and run attacks at scale.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.crowdstrike.com/en-us/blog/vapor-panda-uses-artex-to-scale-global-campaign/",
        "label": "CrowdStrike: China-nexus VAPOR PANDA runs the AI-driven ARTEX framework on DeepSeek to scale a global espionage campaign",
        "date": "2026-10-29"
      },
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold",
        "label": "Google GTIG: actors run open-source multi-agent frameworks that wire commercial and open-weight LLMs to 100+ security tools in live intrusions",
        "date": "2026-11-16"
      }
    ],
    "also_said": [
      {
        "url": "https://danielmiessler.com/blog/why-arent-things-worse",
        "title": "Why Aren’t Things Worse?",
        "date": "2026-08-04"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2085486372337692866",
        "title": "X post: It’s about to get way easier to do many…",
        "date": "2026-08-06"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_ai-models-are-becoming-thanos-guantlets-activity-7491252644712177664-xRP-",
        "title": "LinkedIn post: We are not prepared for a world where every…",
        "date": "2026-08-06"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2112457997926760457",
        "title": "X post: 💯. And as predicted. 1. Better and better open…",
        "date": "2026-10-20"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_expect-to-see-a-whole-lot-more-of-this-activity-7518176874435403776-UsU_",
        "title": "LinkedIn post: Expect to see a whole lot more of this.…",
        "date": "2026-10-20"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2065870441898766426",
        "title": "X post: Yes some can. And some had harnesses long before…",
        "date": "2026-06-13"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_this-weeks-unsupervised-learning-no-activity-7484380757696622592-rLWc",
        "title": "LinkedIn post: AI attack harnesses. Everyone suddenly has one, and they're…",
        "date": "2026-07-18"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_this-weeks-unsupervised-learning-no-activity-7485358114335059968-ENWt",
        "title": "LinkedIn post: AI attack harnesses shipped this week, from Anthropic, Datadog,…",
        "date": "2026-07-21"
      }
    ],
    "authorship": "own-commentary-in-mixed",
    "authorship_note": "Newsletter editions are \"mixed\" in the corpus. This passage sits in Daniel's first-person \"My Analysis\" commentary on Anthropic's AI-orchestrated espionage disclosure; the summarized report text is not quoted."
  },
  {
    "id": "your-ai-versus-their-ai-2025",
    "date": "2025-12-31",
    "quote": "It becomes very clear that the primary security question for a company is how good their attackers' AI is vs. their own … CISOs increasingly realize that there is no way to scale their human team to deal with how constant, continuous, and increasingly effective their attackers are becoming at attacking them",
    "url": "https://danielmiessler.com/blog/cybersecurity-ai-changes-2026",
    "source": "blog",
    "title": "Cybersecurity Changes I Expect in 2026",
    "theme": "consequences",
    "claim": "In 2026 a company's main security question becomes how good its attackers' AI is versus its own, because human teams cannot scale to constant attack.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.crowdstrike.com/en-us/blog/crowdstrike-uncovers-new-agentic-ai-campaigns/",
        "label": "CrowdStrike documents four previously unreported actors running agentic AI intrusion campaigns",
        "date": "2026-10-21"
      },
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold",
        "label": "Google GTIG: agentic threat activity crosses a significant threshold as actors deploy autonomous multi-agent frameworks",
        "date": "2026-11-16"
      }
    ],
    "also_said": [
      {
        "url": "https://www.youtube.com/watch?v=N3J5_JxJRic",
        "title": "Attackers vs. Defenders: The AI Stack Game #shorts",
        "date": "2026-01-26"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_the-primary-game-now-for-defeners-is-your-activity-7421626496638853120-dSer",
        "title": "LinkedIn post: The primary game now for defeners is your AI…",
        "date": "2026-01-26"
      },
      {
        "url": "https://danielmiessler.com/blog/the-great-transition",
        "title": "The Great Transition",
        "date": "2026-02-28"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_in-my-2026-predictions-i-said-the-main-security-activity-7488733815360647168-_B2u",
        "title": "LinkedIn post: In my 2026 predictions, I said the main security…",
        "date": "2026-07-30"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_what-i-expect-to-see-change-in-cybersecurity-activity-7412928165469302784-EgcC",
        "title": "LinkedIn post: What I expect to see change in Cybersecurity in…",
        "date": "2026-01-02"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-511-serious-supabase-and-n8n-vulns-wth-happened-in-december-with-ai-19-chan",
        "title": "Unsupervised Learning NO. 511",
        "date": "2026-01-08"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-530",
        "title": "Unsupervised Learning NO. 530",
        "date": "2026-05-29"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/ul-416",
        "title": "UL NO. 416: Tracking AI Agent Activity, 400 SF Cameras, AI Sleeper Agents...",
        "date": "2024-01-22"
      },
      {
        "url": "https://webtalk.show/episodes/074-are-we-ready-for-claude-mythos-a-cybersecurity-ctos-perspective/",
        "title": "Are We Ready For Claude Mythos? A Cybersecurity CTO's Perspective (The Web Talk Show #74)",
        "date": "2026-04-14"
      },
      {
        "url": "https://www.youtube.com/watch?v=K8h9fEgf5yY",
        "title": "AIMM: A model for thinking about AI and Work",
        "date": "2025-11-20"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "skills-remove-the-hacking-filter-2026",
    "date": "2026-03-27",
    "quote": "isn't thinking about new vulnerabilities, it's stitching up context on the target(s), creating and maintaining tooling, and building and maintaining workflows that allow you to produce quality, consistent tests at scale. Tons of maintenance and support to build this stuff. And it's a massive filter for who can and cannot do the work.",
    "url": "https://danielmiessler.com/blog/ai-unmasked-our-work-as-scaffolding",
    "source": "blog",
    "title": "AI Unmasked Our Work as Scaffolding",
    "theme": "harnesses",
    "claim": "Most hacking is scaffolding (target context, tooling, workflows) that AI skills can package, removing the filter that kept most people from doing it.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://research.openanalysis.net/claude/codex/hacking/ai%20hacking/llm/redteam/policy%20violation/2026/06/16/compromised-claude-hacking.html",
        "label": "OALABS: a low-skill attacker using vague prompts had coding agents breach at least 14 companies",
        "date": "2026-06-16"
      },
      {
        "url": "https://www.anthropic.com/threat-intelligence-report-september-2026",
        "label": "Anthropic: actors run automated 'exploit foundries' and autonomous attack frameworks",
        "date": "2026-09-10"
      }
    ],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/2081900011588305386",
        "title": "X post: The ability to use a command line enough to…",
        "date": "2026-07-28"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_its-remarkable-to-me-that-so-many-people-activity-7528216451900526593-SyEX",
        "title": "LinkedIn post: It's remarkable to me that so many people in…",
        "date": "2026-11-16"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2029680117606453597",
        "title": "X post: The reason people are focusing so much on offensive…",
        "date": "2026-03-05"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2029254928510992573",
        "title": "X post: Penetration testing and bug bounty are two of the…",
        "date": "2026-03-04"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "one-attacker-harness-millions-of-targets-2026",
    "date": "2026-04-12",
    "quote": "But with the crucial difference that the 5 can be spread and scaled across the entire internet/world very quickly. … A 9 attacker in 2019 could only hit very few companies. A level 9 attacker can now/soon build a level 7 AI Harness to smash millions of companies that have level 3 defenses.",
    "url": "https://danielmiessler.com/blog/ai-only-has-to-beat-3-out-of-10",
    "source": "blog",
    "title": "AI Only Has to Beat 3/10",
    "theme": "harnesses",
    "claim": "A skilled attacker can build an AI harness that scales mid-level attack skill across millions of poorly defended companies at once.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.crowdstrike.com/en-us/blog/crowdstrike-uncovers-new-agentic-ai-campaigns/",
        "label": "CrowdStrike documents four previously unreported actors running agentic campaigns with supervisor/worker agent harnesses",
        "date": "2026-10-21"
      },
      {
        "url": "https://www.crowdstrike.com/en-us/blog/vapor-panda-uses-artex-to-scale-global-campaign/",
        "label": "CrowdStrike: VAPOR PANDA uses an AI attack framework to scale a global campaign",
        "date": "2026-10-29"
      }
    ],
    "also_said": [
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_i-think-theres-a-misconception-about-how-activity-7448832336747470848-9PH2",
        "title": "LinkedIn post: A 9 attacker in 2019 could only hit very…",
        "date": "2026-04-11"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2043066324591743251",
        "title": "X post: The Mythos hype has convinced people that AI is…",
        "date": "2026-04-11"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-542",
        "title": "Unsupervised Learning NO. 542",
        "date": "2026-09-10"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2044840342701375899",
        "title": "X post: I think this is correct in terms of public…",
        "date": "2026-04-16"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2098475183736824077",
        "title": "X post: 🧠 Unsupervised Learning NO. 542 is out. 🔓 AI…",
        "date": "2026-09-11"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-533",
        "title": "Unsupervised Learning NO. 533",
        "date": "2026-06-20"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2068793439777673485",
        "title": "X post: Well, they were skilled. Well, they had a harness.…",
        "date": "2026-06-21"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2038630041903612380",
        "title": "X post: Everything talked about was from existing or previous models,…",
        "date": "2026-03-30"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "public-techniques-feed-adversary-ai-stacks-2026",
    "date": "2026-05-31",
    "quote": "Basically any new public technique goes right into your competitors and adversaries’ AI automation stack.",
    "url": "https://x.com/DanielMiessler/status/2061189354475893016",
    "source": "tweet",
    "title": "X post, May 31, 2026",
    "theme": "harnesses",
    "claim": "Every newly published attack technique will be absorbed straight into adversaries' AI automation stacks.",
    "status": "pending",
    "evidence": [],
    "also_said": [],
    "authorship": "own-account",
    "authorship_note": "Posted from Daniel's own X account (x.com/DanielMiessler); corpus attribution is \"unverified\" for all social posts. The quoted words are the post's own text, not a quote of anyone else."
  },
  {
    "id": "open-models-remove-lab-oversight-2026",
    "date": "2026-06-13",
    "quote": "First, it would allow this level of malicious cyber activity without needing to hide from Anthropic / OpenAI. The only constraint would be attacker skill and compute.",
    "url": "https://x.com/DanielMiessler/status/2065809024856244400",
    "source": "tweet",
    "title": "X post, June 13, 2026",
    "theme": "open-models",
    "claim": "Open models at frontier cyber level let attackers work without lab monitoring; only skill and compute will constrain them.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/research/glm-5-3-and-the-spread-of-advanced-cyber-capabilities",
        "label": "Anthropic: open-weight GLM-5.3 builds end-to-end exploits like Mythos Preview, outside any lab's safeguards",
        "date": "2026-09-29"
      },
      {
        "url": "https://www.crowdstrike.com/en-us/blog/unknown-threat-actor-uses-artex-to-target-south-korean-finance/",
        "label": "CrowdStrike: ARTEX intrusions against South Korean finance ran on open DeepSeek and GLM models",
        "date": "2026-10-07"
      }
    ],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/2072388289542422656",
        "title": "X post: Models that can do the bad stuff should be…",
        "date": "2026-07-01"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2072323602628874247",
        "title": "X post on zero controls for open-source models",
        "date": "2026-07-01"
      },
      {
        "url": "https://danielmiessler.com/blog/handguns-in-the-convenience-aisle",
        "title": "Why Don’t We Put Handguns in the Convenience Aisle?",
        "date": "2026-07-03"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2082677761274232835",
        "title": "X post: Attacker don’t need to sneak through any guardrails. They…",
        "date": "2026-07-30"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2085098737542774939",
        "title": "X post: “If OpenAI and Anthropic’s models can do this with…",
        "date": "2026-08-05"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_one-of-the-things-everyone-should-be-thinking-activity-7490864528222547968-uWVj",
        "title": "LinkedIn post: One of the things everyone should be thinking about this Blackhat / DEFCON…",
        "date": "2026-08-05"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2085108097111245058",
        "title": "X post: Clarification: I mean how impressive Sol and Opus and…",
        "date": "2026-08-05"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2086566272050110722",
        "title": "X post: 100%. People have no idea how weird it's about…",
        "date": "2026-08-09"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2081909971848139047",
        "title": "X post: I agree the cat is out of the bag…",
        "date": "2026-07-28"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2082654606438572051",
        "title": "X post: So does that mean anyone in the world should…",
        "date": "2026-07-30"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_that-clip-of-anthropics-ceo-calling-open-source-activity-7478888807371595776-YsgS",
        "title": "LinkedIn post: If ChatGPT, Claude, and the most popular open source…",
        "date": "2026-07-03"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2071223094216413435",
        "title": "X post: I think the calculus for closed models is way…",
        "date": "2026-06-28"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-534",
        "title": "Unsupervised Learning NO. 534",
        "date": "2026-06-25"
      }
    ],
    "authorship": "own-account",
    "authorship_note": "Posted from Daniel's own X account (x.com/DanielMiessler); corpus attribution is \"unverified\" for all social posts. The quoted words are the post's own text, not a quote of anyone else."
  },
  {
    "id": "ai-finds-and-exploits-critical-systems-2026",
    "date": "2026-06-27",
    "quote": "The new AI models are becoming genuinely surprising and concerning in their ability to find and exploit vulnerabilities … The chances of serious compromises of critical systems are very high, and increasing.",
    "url": "https://x.com/DanielMiessler/status/2070662670332678350",
    "source": "tweet",
    "title": "X post, June 27, 2026",
    "theme": "consequences",
    "claim": "New models' skill at finding and exploiting vulnerabilities makes serious compromises of critical systems very likely, and likelier over time.",
    "status": "pending",
    "evidence": [],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/2070657879099510852",
        "title": "X post: The software that runs our critical infrastructure is still,…",
        "date": "2026-06-26"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2070615309875089423",
        "title": "X post: So do you agree some level of thought and…",
        "date": "2026-06-26"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2068799203724386579",
        "title": "X post: What I'm saying is the guy who knows what's…",
        "date": "2026-06-21"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2068429887313232275",
        "title": "X post: How much stability degradation would have to happen from…",
        "date": "2026-06-20"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2070381107279991242",
        "title": "X post: If the labs were just releasing these models, and…",
        "date": "2026-06-26"
      }
    ],
    "authorship": "own-account",
    "authorship_note": "Posted from Daniel's own X account (x.com/DanielMiessler); corpus attribution is \"unverified\" for all social posts. The quoted words are the post's own text, not a quote of anyone else."
  },
  {
    "id": "models-will-plan-attacks-and-evade-detection-2026",
    "date": "2026-07-30",
    "quote": "If you think about where AI models will be in six months or two years, they will have extraordinary ability to manipulate people, plan attacks, help people avoid detection while conducting attacks",
    "url": "https://x.com/DanielMiessler/status/2082688008244093343",
    "source": "tweet",
    "title": "X post, July 30, 2026",
    "theme": "consequences",
    "claim": "Within six months to two years, AI models will be extraordinarily good at planning attacks, manipulating people and helping attackers avoid detection.",
    "status": "pending",
    "evidence": [],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/2092762498521080282",
        "title": "X post: I basically think it’s about to become one of the primary weapons…",
        "date": "2026-08-26"
      }
    ],
    "authorship": "own-account",
    "authorship_note": "Posted from Daniel's own X account (x.com/DanielMiessler); corpus attribution is \"unverified\" for all social posts. The quoted words are the post's own text, not a quote of anyone else."
  },
  {
    "id": "open-models-reach-frontier-attack-level-2026",
    "date": "2026-08-07",
    "quote": "Unrestricted open source models are going to be as good as Sol / Mythos / Astra within 3-12 months, and that's basically going to be a Thanos Guantlet situation.",
    "url": "https://www.linkedin.com/posts/danielmiessler_have-you-ever-thought-about-how-easy-it-would-activity-7491601439564353536-gIms",
    "source": "linkedin",
    "title": "Have you ever thought about how easy it would be to hack you? (LinkedIn)",
    "theme": "open-models",
    "claim": "Within 3-12 months, unrestricted open models will match the best closed models at attack work, putting frontier capability in anyone's hands.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.nist.gov/news-events/news/2026/09/caisis-assessment-zais-glm-53-cyber-capabilities",
        "label": "NIST CAISI: open-weight GLM-5.3 is the most cyber-capable open model to date but still about four months behind the US frontier: close, not equal",
        "date": "2026-09-17"
      },
      {
        "url": "https://www.anthropic.com/research/glm-5-3-and-the-spread-of-advanced-cyber-capabilities",
        "label": "Anthropic: open-weight GLM-5.3 can build end-to-end cyber exploits as Mythos Preview can, though it is not shown to match the best closed models across the board",
        "date": "2026-09-29"
      }
    ],
    "also_said": [
      {
        "url": "https://danielmiessler.com/blog/how-easy-to-hack-you",
        "title": "How Easy It Would Be to Hack You",
        "date": "2026-08-15"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2066641553410650589",
        "title": "X post: What I don’t think people are understanding is that…",
        "date": "2026-06-15"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_what-i-dont-think-people-are-understanding-activity-7472413851515244544-sfyx",
        "title": "LinkedIn post: What I don’t think people are understanding is that…",
        "date": "2026-06-15"
      },
      {
        "url": "https://www.youtube.com/watch?v=a5L-lFd8jSk",
        "title": "Why We Need Some AI Model Controls",
        "date": "2026-07-02"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-540",
        "title": "Unsupervised Learning NO. 540",
        "date": "2026-08-31"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2086591850509709524",
        "title": "X post: Just wait til everyone has a Mythos-level Thanos Guantlet…",
        "date": "2026-08-09"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2066741850388750654",
        "title": "X post: As someone who’s done offensive cyber for decades, I…",
        "date": "2026-06-16"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2041916431470838047",
        "title": "X post: We’re missing a much bigger point on Mythos. It…",
        "date": "2026-04-08"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2041946894990258491",
        "title": "X post: Mythos isn't even a cybersecurity model. It's just good…",
        "date": "2026-04-08"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2044725080182403175",
        "title": "X post: This whole Mythos thing continues to surprise me. Not…",
        "date": "2026-04-16"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_were-missing-a-much-bigger-point-on-mythos-activity-7447682935102697472-9wIw",
        "title": "LinkedIn post: We’re missing a much bigger point on Mythos. It…",
        "date": "2026-04-08"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_this-whole-mythos-thing-continues-to-surprise-activity-7450480336783712256-vNkb",
        "title": "LinkedIn post: This whole Mythos thing continues to surprise me. Not…",
        "date": "2026-04-16"
      },
      {
        "url": "https://danielmiessler.com/blog/wrong-message-from-mythos",
        "title": "We’re Getting the Wrong Message from Mythos",
        "date": "2026-04-07"
      },
      {
        "url": "https://danielmiessler.com/blog/mythos-is-new-normal",
        "title": "Mythos is Just the New Normal",
        "date": "2026-04-15"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-526",
        "title": "Unsupervised Learning NO. 526",
        "date": "2026-04-20"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_%F0%9D%97%A7%F0%9D%97%B5%F0%9D%97%B6%F0%9D%98%80-%F0%9D%98%84%F0%9D%97%B2%F0%9D%97%B2%F0%9D%97%B8%F0%9D%98%80-%F0%9D%97%A8%F0%9D%97%BB%F0%9D%98%80%F0%9D%98%82%F0%9D%97%BD%F0%9D%97%B2%F0%9D%97%BF%F0%9D%98%83%F0%9D%97%B6%F0%9D%98%80%F0%9D%97%B2%F0%9D%97%B1-activity-7500290342672465920-zfpu",
        "title": "LinkedIn post: unrestricted open models are about to put frontier-level attack…",
        "date": "2026-08-31"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_why-we-need-some-ai-model-controls-activity-7478499400642883584-sVJG",
        "title": "LinkedIn post: My argument for why we need (some) limited controls…",
        "date": "2026-07-02"
      }
    ],
    "authorship": "own-account",
    "authorship_note": "Posted from Daniel's own LinkedIn account (linkedin.com/posts/danielmiessler_); his blog post 'How Easy It Would Be to Hack You' repeats it. \"Guantlet\" is his spelling in the post. Corpus attribution is \"unverified\" for all social posts."
  },
  {
    "id": "better-ai-faster-compromise-2026",
    "date": "2026-08-17",
    "quote": "the better general AI gets … the faster your internet-facing mistakes will get compromised.",
    "url": "https://danielmiessler.com/blog/how-ai-builders-get-hacked",
    "source": "blog",
    "title": "How AI Builders Will Get Hacked",
    "theme": "consequences",
    "claim": "As general AI improves, anything you leave exposed online will be found and compromised faster.",
    "status": "happening",
    "evidence": [
      {
        "url": "https://www.anthropic.com/threat-intelligence-report-september-2026",
        "label": "Anthropic: actors run round-the-clock AI vulnerability research and exploit writing in automated 'exploit foundries'",
        "date": "2026-09-10"
      },
      {
        "url": "https://cloud.google.com/blog/topics/threat-intelligence/agentic-threat-activity-crosses-a-significant-threshold",
        "label": "Google GTIG: agentic frameworks execute multi-stage intrusions at greater operational velocity",
        "date": "2026-11-16"
      }
    ],
    "also_said": [
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-538",
        "title": "Unsupervised Learning NO. 538",
        "date": "2026-07-30"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  },
  {
    "id": "open-models-outpace-injection-defenses-2026",
    "date": "2026-08-19",
    "quote": "The most interesting and concerning part of this to me is that this is a game of the strength of prompt injection defenses versus the rapidly increasing intelligence of unrestricted open source models. And I don't like the odds for us in this fight.",
    "url": "https://www.linkedin.com/posts/danielmiessler_im-worried-about-a-prompt-injection-worm-activity-7495928850309099521-QMpQ",
    "source": "linkedin",
    "title": "I'm Worried About a Prompt Injection Worm (LinkedIn)",
    "theme": "combination",
    "claim": "A self-spreading prompt-injection worm built with unrestricted open models could hit the AI agents parsing everyone's email, and defenses will likely lose.",
    "status": "pending",
    "evidence": [],
    "also_said": [
      {
        "url": "https://danielmiessler.com/blog/prompt-injection-worm",
        "title": "I’m Worried About a Prompt Injection Worm",
        "date": "2026-08-18"
      }
    ],
    "authorship": "own-account",
    "authorship_note": "Posted from Daniel's own LinkedIn account (linkedin.com/posts/danielmiessler_); the same words are in his blog post of 2026-08-18, where a markdown link splits the sentence. Corpus attribution is \"unverified\" for all social posts."
  },
  {
    "id": "bureaucracy-hands-attackers-the-edge-2026",
    "date": "2026-09-22",
    "quote": "Attackers will be doing this in minutes, or hours, while for most businesses, and for the foreseeable future, that will still take days or weeks or months. Hell, even years. … This is not a fair fight in the slightest, and the root cause is bureaucratic friction, not technology.",
    "url": "https://danielmiessler.com/blog/attacker-defender-ai-advantage",
    "source": "blog",
    "title": "Attacker vs. Defender AI Advantage",
    "theme": "consequences",
    "claim": "Bureaucracy will give attackers the AI edge: they will adopt new capabilities in minutes or hours while companies take months.",
    "status": "pending",
    "evidence": [],
    "also_said": [
      {
        "url": "https://x.com/DanielMiessler/status/2102859075449139319",
        "title": "X post: I think the main AI disadvantage defenders have vs.…",
        "date": "2026-09-23"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_attacker-vs-defender-ai-advantage-activity-7508627213978869760-mwfN",
        "title": "LinkedIn post: Attackers can learn and adjust in minutes or hours,…",
        "date": "2026-09-23"
      },
      {
        "url": "https://www.linkedin.com/posts/danielmiessler_when-you-figure-out-why-just-patch-is-shitty-activity-7527889713509072896-cOwN",
        "title": "LinkedIn post: When you figure out why “just patch” is shitty…",
        "date": "2026-11-16"
      },
      {
        "url": "https://newsletter.danielmiessler.com/p/unsupervised-learning-no-544",
        "title": "Unsupervised Learning NO. 544",
        "date": "2026-09-23"
      },
      {
        "url": "https://x.com/DanielMiessler/status/2102607048588624370",
        "title": "X post sharing “Attacker vs. Defender AI Advantage”",
        "date": "2026-09-23"
      }
    ],
    "authorship": "principal",
    "authorship_note": "Corpus marks this danielmiessler.com document principal-authored."
  }
]
