{
 "public_id": "AIR-2025-0069",
 "incident_id": "2025-07-llm-share-link-conversation-indexing",
 "date_disclosed": "2025-07-31",
 "date_occurred": "2024",
 "title": "Over 100,000 shared ChatGPT, Claude, Copilot, Qwen, Mistral and Grok conversations were indexed by search engines and permanently archived",
 "class": "in_the_wild",
 "attack_surface": "memory_store",
 "initial_vector": "none_autonomous_action",
 "agent_capability_abused": [
  "conversation_sharing",
  "public_link_generation"
 ],
 "target_product": "share features across multiple LLM platforms",
 "target_vendor": "OpenAI, Anthropic, Microsoft, Alibaba, Mistral, xAI",
 "autonomy_level": "assistant",
 "impact_realized": true,
 "impact_type": [
  "privacy_violation",
  "information_disclosure",
  "credential_exposure"
 ],
 "blast_radius": "over 100,000 conversations indexed; roughly 110,000 preserved in the Wayback Machine even after de-indexing, containing API keys, access tokens, personal identifiers and business data",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI03"
  ],
  "mitre_atlas": [],
  "labels": [
   "dark_pattern_default",
   "irreversible_exposure",
   "archived_after_deletion",
   "credential_exposure",
   "cross_vendor"
  ]
 },
 "defense_present": "a 'discoverable' opt-in checkbox users did not understand",
 "defense_bypassed": false,
 "fix_shipped": "OpenAI removed the discoverability option and scrubbed ~50,000 conversations from Google's index — but archived copies persisted",
 "primary_sources": [
  "https://www.digitaldigging.org/p/chatgpt-confessions-gone-they-are",
  "https://incidentdatabase.ai/cite/1186"
 ],
 "verified_quote": "When users click \"share\" on a ChatGPT conversation, they think they're creating a temporary link for a friend or colleague. What they don't realize is they were also creating a permanent, searchable record of their thoughts, confessions, and sometimes illegal activities seen by Archive.org.",
 "confidence": "high",
 "notes": "No attacker and no bug — an interface default did all the work, and the mental model of 'share a link with a colleague' did not match 'publish to the indexable web'. The instructive part is the remediation gap: OpenAI could de-index but could not un-archive, so ~110,000 conversations survived the fix. Investigation by Digital Digging with researcher Nicolas Deleur. AI Incident Database incident 1186.",
 "risk_category": "security_fraud",
 "risk_secondary": [
  "privacy",
  "brand_trust"
 ],
 "risk_subcategory": "Credential & Secret Theft",
 "risk_sub_secondary": [],
 "ai_entity": "generative",
 "agency": "ai_as_target",
 "agency_debatable": true,
 "agency_reason": "a sharing default plus crawler indexing exposed the conversations",
 "vector_family": "none_autonomous",
 "vector_detail": "none_autonomous_action",
 "guardrail_kind": [
  "unknown"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [
   1186
  ],
  "atlas": [],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI03",
 "agent_framework": "unknown",
 "model_provider": "multiple",
 "model_name": "",
 "detection_mechanism": "unknown",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "credential_theft",
 "eai_domain": "access_control_and_permissions",
 "eai_secondary": [
  "sensitive_data_exposure"
 ],
 "eai_reason": "subcategory: Credential & Secret Theft",
 "asi_curated": [],
 "asi_scope": "in_scope",
 "asi_provenance": "rule",
 "asi_reason": "ASI03 (eai_category=credential_theft)",
 "submitted_by": "EAI Internal"
}