{
 "public_id": "AIR-2026-0124",
 "incident_id": "2026-06-copilot-cowork-deepseek-optout-bypass",
 "date_disclosed": "2026-06",
 "date_occurred": null,
 "title": "Microsoft Copilot Cowork Skills call DeepSeek and Mistral even when the tenant never opted in, and the Skill scanner rates the Skill 100/100",
 "class": "responsible_disclosure",
 "attack_surface": "skill_plugin",
 "initial_vector": "skill_calls_unapproved_external_model",
 "agent_capability_abused": [
  "skill_execution",
  "external_model_api_call",
  "system_prompt_override"
 ],
 "target_product": "Microsoft Copilot Cowork",
 "target_vendor": "Microsoft",
 "autonomy_level": "semi_autonomous",
 "impact_realized": false,
 "impact_type": [
  "data_exposure",
  "governance_bypass"
 ],
 "blast_radius": "M365 tenants that believe they have blocked DeepSeek/Mistral via admin settings",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI05",
   "ASI01",
   "ASI02"
  ],
  "mitre_atlas": [],
  "labels": [
   "governance_bypass",
   "shadow_model_use",
   "guardrail_bypass",
   "scanner_false_negative",
   "data_residency"
  ]
 },
 "defense_present": "M365 Admin Center opt-in control for DeepSeek Preview; Skill scanner",
 "defense_bypassed": true,
 "fix_shipped": "incomplete - Microsoft removed the DeepSeek setting but Skills retain access; only remaining control is disabling Copilot Cowork",
 "primary_sources": [
  "https://www.promptarmor.com/resources/skill-accesses-deepseek-despite-copilot-cowork-admin-opt-out"
 ],
 "verified_quote": "Skills can query DeepSeek in Microsoft Copilot Cowork, even when your organization has NOT opted in to the DeepSeek Preview.",
 "confidence": "high",
 "notes": "PROVENANCE CAVEAT: no clear publication date on the page; the cited summary rendered a reference date as '2024', which is inconsistent with Copilot Cowork's existence - 2026-06 in the incident_id is provisional and date_disclosed is left null. PromptArmor. Also notes custom system prompts on these calls bypass Copilot's guardrails, and Microsoft's scanner gave the DeepSeek-calling Skill a perfect 100 'low risk' score. | date_disclosed is recorded to the month only; no source establishes the exact day.",
 "risk_category": "privacy",
 "risk_secondary": [
  "product_misuse"
 ],
 "risk_subcategory": "System Prompt Extraction",
 "risk_sub_secondary": [],
 "ai_entity": "generative",
 "agency": "agent_acted",
 "agency_debatable": false,
 "agency_reason": "the Skills called unapproved external models with tenant data",
 "vector_family": "other",
 "vector_detail": "skill_calls_unapproved_external_model",
 "guardrail_kind": [
  "unknown"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [],
  "atlas": [],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI05",
 "agent_framework": "unknown",
 "model_provider": "multiple",
 "model_name": "",
 "detection_mechanism": "external_researcher",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "sensitive_data_exposure",
 "eai_domain": "privacy",
 "eai_secondary": [],
 "eai_reason": "subcategory: System Prompt Extraction",
 "asi_curated": [
  "ASI05"
 ],
 "asi_scope": "in_scope",
 "asi_provenance": "curated_plus_rule",
 "asi_reason": "curated: ASI05 | added ASI01 (vector=injection), ASI02 (impact=tool_misuse)",
 "submitted_by": "EAI Internal"
}