{
 "as_of": "2026-10-25",
 "source": "Compiled by wheresthe.ai from lab disclosures, regulator and press records",
 "count": 27,
 "records": [
  {
   "id": "us-johns-hopkins-bayview-patient-data-put-into-public-ai-tool-2026-10",
   "system": "an unapproved public AI tool (unnamed)",
   "operator": "a Johns Hopkins Bayview Medical Center employee",
   "type": "personal health information of 3,182 patients uploaded",
   "disclosed_by": "Johns Hopkins Bayview Medical Center, notifying patients",
   "status": "patients notified, reported 24 Oct 2026",
   "date": "2026-10-24",
   "jurisdictions": [
    "US"
   ],
   "title": "Johns Hopkins Bayview tells 3,182 patients an employee put their health data into a public AI tool",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0424",
   "story": "https://wheresthe.ai/d/us-johns-hopkins-bayview-patient-data-put-into-public-ai-tool-2026-10/",
   "source": "https://www.baltimoresun.com/2026/10/24/johns-hopkins-bayview-medical-center-data-breach/"
  },
  {
   "id": "intl-attackers-use-ai-agents-after-php2shell-exploits-2026-10",
   "system": "agentic AI attack platforms (unnamed)",
   "operator": "unknown attackers exploiting the PHP2Shell flaws",
   "type": "post-exploitation of a European government entity: password hashes, MFA secrets, NTLM relay into Active Directory",
   "disclosed_by": "Palo Alto Networks Unit 42",
   "status": "reported 23 Oct 2026; Domain Admin escalation attempted",
   "date": "2026-10-23",
   "jurisdictions": [
    "INTL"
   ],
   "title": "Attackers exploiting the PHP flaws hand the break-in to AI agents, Unit 42 says",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0423",
   "story": "https://wheresthe.ai/d/intl-attackers-use-ai-agents-after-php2shell-exploits-2026-10/",
   "source": "https://unit42.paloaltonetworks.com/cve-2026-109618-cve-2026-109619/"
  },
  {
   "id": "intl-ransomware-affiliates-ran-intrusions-through-ai-agents-2026-10",
   "system": "commercial models on OpenRouter, driven by the Phantom framework through OpenCode",
   "operator": "affiliates of the INC and The Gentlemen ransomware groups",
   "type": "AI-executed, human-approved intrusions and data extortion",
   "disclosed_by": "SOCRadar",
   "status": "disclosed",
   "date": "2026-10-23",
   "jurisdictions": [
    "INTL"
   ],
   "title": "Ransomware affiliates ran intrusions through AI agents, an exposed server shows",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0396",
   "story": "https://wheresthe.ai/d/intl-ransomware-affiliates-ran-intrusions-through-ai-agents-2026-10/",
   "source": "https://thehackernews.com/2026/10/exposed-infrastructure-reveals-how.html"
  },
  {
   "id": "es-spanish-opposition-party-launches-deepfake-video-in-elect-2026-10",
   "system": "AI-generated deepfake video",
   "operator": "People’s Party (Spain)",
   "type": "election video with deepfakes of people linked to the Socialist party, carrying a nearly transparent AI note",
   "disclosed_by": "The Guardian",
   "status": "published; nearly 2 million views",
   "date": "2026-10-23",
   "jurisdictions": [
    "ES"
   ],
   "title": "Spanish opposition party launches deepfake video in election campaign",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0383",
   "story": "https://wheresthe.ai/d/es-spanish-opposition-party-launches-deepfake-video-in-elect-2026-10/",
   "source": "https://www.theguardian.com/world/2026/oct/23/opposition-deepfake-video-spain-election"
  },
  {
   "id": "intl-ai-found-php-flaws-take-websites-offline-2026-10",
   "system": "OpenAI GPT-5.6 Sol and Google Gemini 4 Argon, used to find the flaws",
   "operator": "Calif (OpenAI-funded audit of PHP) and Wiz",
   "type": "two critical PHP flaws letting an unauthenticated request run code; websites taken offline as a precaution",
   "disclosed_by": "PHP security team; details leaked a day before the fix",
   "status": "fixed on 21 Oct 2026; widespread exploitation attempts reported",
   "date": "2026-10-21",
   "jurisdictions": [
    "INTL",
    "FR",
    "AU",
    "SG"
   ],
   "title": "AI-found PHP flaws push governments and companies to take websites offline",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0372",
   "story": "https://wheresthe.ai/d/intl-ai-found-php-flaws-take-websites-offline-2026-10/",
   "source": "https://www.thehindu.com/sci-tech/technology/websites-go-dark-as-ai-exposes-critical-security-flaws/article71609942.ece"
  },
  {
   "id": "us-openai-says-two-ai-agents-weakened-parts-of-its-internal-2026-10",
   "system": "GPT-6.1 Sol and GPT-6 Astra coding agents (internal deployments)",
   "operator": "OpenAI",
   "type": "misalignment monitoring silenced (3 Oct) and removed (6 Oct) during internal coding tasks",
   "disclosed_by": "OpenAI",
   "status": "disclosed; Auto-Review now blocks disabling monitoring; cause under investigation",
   "date": "2026-10-20",
   "jurisdictions": [
    "US"
   ],
   "title": "OpenAI reports two agents that silenced or removed its misalignment monitoring",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0347",
   "story": "https://wheresthe.ai/d/us-openai-says-two-ai-agents-weakened-parts-of-its-internal-2026-10/",
   "source": "https://alignment.openai.com/misalignment-reports/suppressing-misalignment-monitoring-alerts-and-disabling-misalignment-monitoring/"
  },
  {
   "id": "us-chinese-speaking-hacker-used-ai-agents-to-break-into-site-2026-10",
   "system": "AI agents on the Cairn platform (DeepSeek and Claude Code models)",
   "operator": "Chinese-speaking hacker",
   "type": "cyber attack on more than 100 company websites",
   "disclosed_by": "NorForge; Reuters",
   "status": "disclosed; Deere, SharkNinja and Cummins responded",
   "date": "2026-10-20",
   "jurisdictions": [
    "US",
    "CA",
    "AU",
    "JP"
   ],
   "title": "Chinese-speaking hacker used AI agents to break into sites of 100+ companies for about US$5000",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0336",
   "story": "https://wheresthe.ai/d/us-chinese-speaking-hacker-used-ai-agents-to-break-into-site-2026-10/",
   "source": "https://www.itnews.com.au/news/how-one-hacker-used-ai-agents-and-a-few-thousand-dollars-to-break-into-scores-of-sites-629764"
  },
  {
   "id": "jp-china-made-ai-tool-artex-used-in-cyberattacks-on-japanese-2026-10",
   "system": "ARTEX (autonomous AI penetration-testing tool)",
   "operator": "unidentified attack group",
   "type": "data breaches at Japanese companies (about 20 targeted by one host; eight had disclosed leaks)",
   "disclosed_by": "GMO Cybersecurity by Ierae",
   "status": "shared with law enforcement; investigation continues",
   "date": "2026-10-19",
   "jurisdictions": [
    "JP",
    "KR",
    "CN"
   ],
   "title": "GMO confirms the AI penetration-testing tool ARTEX behind attacks on Japanese companies",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0344",
   "story": "https://wheresthe.ai/d/jp-china-made-ai-tool-artex-used-in-cyberattacks-on-japanese-2026-10/",
   "source": "https://gmo-cybersecurity.com/blog/20261019/"
  },
  {
   "id": "kp-ai-guided-torpedo-test-2026-10",
   "system": "AI-guided torpedo",
   "operator": "North Korean military",
   "type": "weapon test",
   "disclosed_by": "KCNA (state media)",
   "status": "claimed success",
   "date": "2026-10-17",
   "jurisdictions": [
    "KP"
   ],
   "title": "North Korea says it tested a torpedo that uses AI for target detection and guidance",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0318",
   "story": "https://wheresthe.ai/d/kp-ai-guided-torpedo-test-2026-10/",
   "source": "https://www.scmp.com/news/asia/east-asia/article/3371218/north-korea-says-it-tested-new-torpedo-ai-tech"
  },
  {
   "id": "cn-ai-agents-on-tencent-cloud-made-more-than-2-600-attempts-2026-10",
   "system": "AI agents on Tencent Cloud (third-party models under test)",
   "operator": "Tencent AI evaluation team",
   "type": "unsanctioned agent actions",
   "disclosed_by": "swarmchasers (researchers)",
   "status": "Tencent strengthening access controls",
   "date": "2026-10-16",
   "jurisdictions": [
    "CN"
   ],
   "title": "AI agents on Tencent Cloud made more than 2,600 attempts to extract data from Alibaba’s Amap",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0305",
   "story": "https://wheresthe.ai/d/cn-ai-agents-on-tencent-cloud-made-more-than-2-600-attempts-2026-10/",
   "source": "https://www.scmp.com/tech/tech-trends/article/3371007/ai-agents-tencent-cloud-try-break-alibabas-amap-researchers"
  },
  {
   "id": "us-anthropic-says-new-defences-cut-distillation-of-claude-by-up-to-99-2026-10",
   "system": "Claude (target of distillation)",
   "operator": "distillers using networks of fraudulent accounts; Zhipu named",
   "type": "industrial-scale distillation of model outputs",
   "disclosed_by": "Anthropic",
   "status": "defences in place; harvested data down about 90 to 99 per cent",
   "date": "2026-10-12",
   "jurisdictions": [
    "US",
    "CN"
   ],
   "title": "Anthropic says new defences cut the data distillers take from Claude by about 90 to 99 per cent",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0409",
   "story": "https://wheresthe.ai/d/us-anthropic-says-new-defences-cut-distillation-of-claude-by-up-to-99-2026-10/",
   "source": "https://www.anthropic.com/news/distillation-defenses"
  },
  {
   "id": "us-pa-anthropic-ai-model-submitted-false-murder-tip-to-phila-2026-10",
   "system": "Claude (test run)",
   "operator": "Anthropic",
   "type": "unintended model action",
   "disclosed_by": "Philadelphia Police Department; Anthropic",
   "status": "disclosed",
   "date": "2026-10-09",
   "jurisdictions": [
    "US-PA"
   ],
   "title": "Anthropic AI model submitted false murder tip to Philadelphia police",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0242",
   "story": "https://wheresthe.ai/d/us-pa-anthropic-ai-model-submitted-false-murder-tip-to-phila-2026-10/",
   "source": "https://www.scmp.com/news/world/united-states-canada/article/3370414/anthropic-ai-model-sent-fake-murder-tip-us-police-hit-government-sites"
  },
  {
   "id": "openai-false-front-influence-operations-2026-10",
   "system": "OpenAI models",
   "operator": "Iranian and Russian influence operations",
   "type": "disinformation",
   "disclosed_by": "OpenAI",
   "status": "accounts banned",
   "date": "2026-10-08",
   "jurisdictions": [
    "US",
    "RU",
    "IR",
    "INTL"
   ],
   "title": "OpenAI bans Russian and Iranian influence operations run through fake journalists",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0149",
   "story": "https://wheresthe.ai/d/openai-false-front-influence-operations-2026-10/",
   "source": "https://openai.com/index/disrupting-ai-enabled-false-front-operations/"
  },
  {
   "id": "kr-bank-breaches-suspected-ai-attacks-2026-10",
   "system": "Unidentified AI hacking tool",
   "operator": "Unknown attackers",
   "type": "cyber attack",
   "disclosed_by": "Financial Services Commission",
   "status": "under investigation",
   "date": "2026-10-04",
   "jurisdictions": [
    "KR"
   ],
   "title": "Korea’s financial regulator calls emergency meeting over suspected AI-driven bank breaches",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0142",
   "story": "https://wheresthe.ai/d/kr-bank-breaches-suspected-ai-attacks-2026-10/",
   "source": "https://www.koreajoongangdaily.com/business/from-banks-to-lenders-suspected-ai-hacks-expose-cracks-in-koreas-financial-defenses/12904281"
  },
  {
   "id": "openai-moonshot-distillation-campaign-2026-09",
   "system": "OpenAI models",
   "operator": "Individuals linked to Moonshot AI (alleged)",
   "type": "misuse: distillation",
   "disclosed_by": "OpenAI",
   "status": "disrupted",
   "date": "2026-09-30",
   "jurisdictions": [
    "US",
    "CN"
   ],
   "title": "OpenAI ties large reasoning-distillation campaign to people linked to Moonshot AI",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0096",
   "story": "https://wheresthe.ai/d/openai-moonshot-distillation-campaign-2026-09/",
   "source": "https://openai.com/index/disrupting-a-coordinated-model-distillation-campaign/"
  },
  {
   "id": "au-openai-models-accessed-government-systems-2026-09",
   "system": "OpenAI models (training and evaluation)",
   "operator": "OpenAI",
   "type": "unauthorised access",
   "disclosed_by": "OpenAI",
   "status": "disclosed; subpoena followed",
   "date": "2026-09-28",
   "jurisdictions": [
    "AU"
   ],
   "title": "OpenAI says its models accessed Australian government systems during training",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0134",
   "story": "https://wheresthe.ai/d/au-openai-models-accessed-government-systems-2026-09/",
   "source": "https://openai.com/index/how-we-will-do-better-for-australia/"
  },
  {
   "id": "gb-aisi-mythos-5-unsanctioned-actions-2026-08",
   "system": "Claude Mythos 5; GPT-5.6 Sol",
   "operator": "Anthropic; OpenAI",
   "type": "unsanctioned agent actions",
   "disclosed_by": "UK AI Security Institute",
   "status": "disclosed",
   "date": "2026-08-04",
   "jurisdictions": [
    "GB"
   ],
   "title": "UK AISI says AI agents took 19 unsanctioned actions online during cyber testing",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0180",
   "story": "https://wheresthe.ai/d/gb-aisi-mythos-5-unsanctioned-actions-2026-08/",
   "source": "https://www.aisi.gov.uk/blog/incident-report-unsanctioned-agent-behaviour-during-cyber-testing"
  },
  {
   "id": "anthropic-claude-cyber-eval-incidents-2026-07",
   "system": "Claude models",
   "operator": "Anthropic",
   "type": "unauthorised access during evaluation",
   "disclosed_by": "Anthropic",
   "status": "disclosed; METR investigation",
   "date": "2026-07-30",
   "jurisdictions": [
    "US"
   ],
   "title": "Anthropic finds three cases of Claude breaking into real systems during cyber evaluations",
   "rule_ids": [],
   "wta_id": "WTA-N-2026-0178",
   "story": "https://wheresthe.ai/d/anthropic-claude-cyber-eval-incidents-2026-07/",
   "source": "https://www.anthropic.com/news/investigating-incidents-cybersecurity-evals"
  },
  {
   "id": "intl-openai-agents-breach-hugging-face-2026-07",
   "system": "GPT-5.6 Sol; unreleased research model",
   "operator": "OpenAI",
   "type": "unauthorised access during evaluation",
   "disclosed_by": "Hugging Face; OpenAI",
   "status": "disclosed; RL training paused",
   "date": "2026-07-21",
   "jurisdictions": [
    "US",
    "INTL"
   ],
   "title": "OpenAI says its models escaped an eval sandbox and breached Hugging Face",
   "rule_ids": [
    "us-ca-sb53-tfaia-2025"
   ],
   "wta_id": "WTA-N-2026-0067",
   "story": "https://wheresthe.ai/d/intl-openai-agents-breach-hugging-face-2026-07/",
   "source": "https://openai.com/index/hugging-face-model-evaluation-security-incident/"
  },
  {
   "id": "anthropic-ai-orchestrated-cyber-espionage-2025-11",
   "system": "Claude Code (jailbroken)",
   "operator": "State-sponsored group (assessed Chinese)",
   "type": "misuse: cyber espionage",
   "disclosed_by": "Anthropic",
   "status": "accounts banned",
   "date": "2025-11-13",
   "jurisdictions": [
    "US",
    "CN"
   ],
   "title": "Anthropic discloses largely AI-run espionage campaign by Chinese state group",
   "rule_ids": [],
   "wta_id": "WTA-N-2025-0047",
   "story": "https://wheresthe.ai/d/anthropic-ai-orchestrated-cyber-espionage-2025-11/",
   "source": "https://www.anthropic.com/news/disrupting-AI-espionage"
  },
  {
   "id": "il-microsoft-disables-imod-unit-azure-ai-2025-09",
   "system": "Azure storage and AI services",
   "operator": "Unit of Israel’s Ministry of Defense",
   "type": "military use",
   "disclosed_by": "Microsoft",
   "status": "services disabled",
   "date": "2025-09-25",
   "jurisdictions": [
    "IL",
    "US"
   ],
   "title": "Microsoft cuts an Israeli defence unit’s access to Azure storage and AI services",
   "rule_ids": [],
   "wta_id": "WTA-N-2025-0035",
   "story": "https://wheresthe.ai/d/il-microsoft-disables-imod-unit-azure-ai-2025-09/",
   "source": "https://blogs.microsoft.com/on-the-issues/2025/09/25/update-on-ongoing-microsoft-review/"
  },
  {
   "id": "hk-arup-deepfake-cfo-video-fraud-2024-05",
   "system": "Deepfake video call",
   "operator": "Unknown fraudsters",
   "type": "fraud",
   "disclosed_by": "Arup; Hong Kong Police",
   "status": "under investigation",
   "date": "2024-05-17",
   "jurisdictions": [
    "HK",
    "GB"
   ],
   "title": "Arup named as victim of HK$200m deepfake video-call fraud in Hong Kong",
   "rule_ids": [],
   "wta_id": "WTA-N-2024-0099",
   "story": "https://wheresthe.ai/d/hk-arup-deepfake-cfo-video-fraud-2024-05/",
   "source": "https://scmp.com/news/hong-kong/law-and-crime/article/3263151/uk-multinational-arup-confirmed-victim-hk200-million-deepfake-scam-used-digital-version-cfo-dupe"
  },
  {
   "id": "il-gospel-ai-targeting-report-2023-11",
   "system": "Habsora targeting system",
   "operator": "Israel Defense Forces",
   "type": "military use",
   "disclosed_by": "+972 Magazine; Local Call",
   "status": "reported",
   "date": "2023-11-30",
   "jurisdictions": [
    "IL"
   ],
   "title": "Israeli outlet +972 reports the IDF’s Gospel AI system generates bombing targets in Gaza",
   "rule_ids": [],
   "wta_id": "WTA-N-2023-0119",
   "story": "https://wheresthe.ai/d/il-gospel-ai-targeting-report-2023-11/",
   "source": "https://www.972mag.com/mass-assassination-factory-israel-calculated-bombing-gaza/"
  },
  {
   "id": "us-unitedhealth-navihealth-nh-predict-lawsuit-2023-11",
   "system": "nH Predict",
   "operator": "UnitedHealth; naviHealth",
   "type": "harm: coverage denial (alleged)",
   "disclosed_by": "Class action complaint",
   "status": "open",
   "date": "2023-11-14",
   "jurisdictions": [
    "US"
   ],
   "title": "Families sue UnitedHealth over nH Predict algorithm used in Medicare Advantage denials",
   "rule_ids": [],
   "wta_id": "WTA-N-2023-0145",
   "story": "https://wheresthe.ai/d/us-unitedhealth-navihealth-nh-predict-lawsuit-2023-11/",
   "source": "https://www.statnews.com/2023/11/14/unitedhealth-class-action-lawsuit-algorithm-medicare-advantage/"
  },
  {
   "id": "sk-election-deepfake-audio-simecka-2023-09",
   "system": "AI-generated audio",
   "operator": "Unknown",
   "type": "election deepfake",
   "disclosed_by": "AFP fact-checkers; Denník N",
   "status": "criminal complaint filed",
   "date": "2023-09-28",
   "jurisdictions": [
    "SK"
   ],
   "title": "Fake audio of Slovak party leader discussing vote rigging spreads before election",
   "rule_ids": [],
   "wta_id": "WTA-N-2023-0099",
   "story": "https://wheresthe.ai/d/sk-election-deepfake-audio-simecka-2023-09/",
   "source": "https://ipi.media/slovakia-deepfake-audio-of-dennik-n-journalist-offers-worrying-example-of-ai-abuse/"
  },
  {
   "id": "es-almendralejo-ai-nude-images-minors-2023-09",
   "system": "AI image generation",
   "operator": "Minors aged 12 to 14",
   "type": "non-consensual images of minors",
   "disclosed_by": "Policía Nacional",
   "status": "under investigation",
   "date": "2023-09-20",
   "jurisdictions": [
    "ES"
   ],
   "title": "Police in Almendralejo, Spain, investigate AI-made nude images of schoolgirls",
   "rule_ids": [],
   "wta_id": "WTA-N-2023-0095",
   "story": "https://wheresthe.ai/d/es-almendralejo-ai-nude-images-minors-2023-09/",
   "source": "https://www.eldiario.es/andalucia/identificados-deepfakes-sexuales-almendralejo-acumulan-22-denuncias-12-14-anos_1_10531559.html"
  },
  {
   "id": "us-fake-pentagon-explosion-image-2023-05",
   "system": "AI-generated image",
   "operator": "Unknown",
   "type": "disinformation",
   "disclosed_by": "Arlington County Fire Department",
   "status": "debunked",
   "date": "2023-05-22",
   "jurisdictions": [
    "US"
   ],
   "title": "Fake, apparently AI-made image of a Pentagon explosion spreads on Twitter as stocks briefly dip",
   "rule_ids": [],
   "wta_id": "WTA-N-2023-0060",
   "story": "https://wheresthe.ai/d/us-fake-pentagon-explosion-image-2023-05/",
   "source": "https://www.snopes.com/fact-check/pentagon-explosion-photo-twitter/"
  }
 ],
 "note": "Informational, not legal advice. Every record links to its official source; check it before relying on a record. Cite as \"Source: wheresthe.ai\"."
}