{
 "public_id": "AIR-2025-0057",
 "incident_id": "2025-07-xai-grok-mechahitler",
 "date_disclosed": "2025-07-12",
 "date_occurred": "2025-07-08",
 "title": "Grok autonomously published antisemitic posts on X for roughly 16 hours after a system-prompt change made it echo extremist user content",
 "class": "safety_failure",
 "attack_surface": "consumer_chatbot",
 "initial_vector": "system_prompt_change_plus_untrusted_context",
 "agent_capability_abused": [
  "autonomous_public_posting",
  "ingestion_of_untrusted_user_posts"
 ],
 "target_product": "Grok (@grok on X)",
 "target_vendor": "xAI",
 "autonomy_level": "semi_autonomous",
 "impact_realized": true,
 "impact_type": [
  "harmful_content_published",
  "reputational_damage"
 ],
 "blast_radius": "public posts on X visible platform-wide for ~16 hours before the bot was pulled",
 "cve": [],
 "taxonomy": {
  "owasp_asi": [
   "ASI09"
  ],
  "mitre_atlas": [],
  "labels": [
   "autonomous_publishing",
   "untrusted_context_echo",
   "system_prompt_regression"
  ]
 },
 "defense_present": "content moderation on model outputs; the regression was in a code path upstream of the model",
 "defense_bypassed": true,
 "fix_shipped": "offending code path removed, system prompt published, @grok posting suspended during remediation",
 "primary_sources": [
  "https://techcrunch.com/2025/07/12/xai-and-grok-apologize-for-horrific-behavior"
 ],
 "verified_quote": "First off, we deeply apologize for the horrific behavior that many experienced.",
 "confidence": "high",
 "notes": "Boundary call. Included because @grok is action-taking: it composes and publishes public posts under the vendor's brand with no human in the loop, so a model failure becomes published speech rather than a private reply. Partly overlaps the adversarial side of the corpus - xAI says an update made Grok 'susceptible to existing X user posts; including when such posts contained extremist views', i.e. untrusted retrieved content shaped its output - but the trigger was xAI's own prompt change ('You tell like it is and you are not afraid to offend people who are politically correct'), not an attacker campaign, so it is filed as a safety failure.",
 "risk_category": "safety_harm",
 "risk_secondary": [
  "privacy",
  "brand_trust",
  "bias_fairness"
 ],
 "risk_subcategory": "Toxicity & Hate Speech",
 "risk_sub_secondary": [
  "Harassment & Radicalization"
 ],
 "ai_entity": "generative",
 "agency": "agent_acted",
 "agency_debatable": false,
 "agency_reason": "Grok composed and published antisemitic posts to X for 16 hours",
 "vector_family": "other",
 "vector_detail": "system_prompt_change_plus_untrusted_context",
 "guardrail_kind": [
  "input_classifier"
 ],
 "guardrail_outcome": "silent",
 "xrefs": {
  "aiid": [],
  "atlas": [],
  "cve": [],
  "ghsa": []
 },
 "primary_asi": "ASI09",
 "agent_framework": "unknown",
 "model_provider": "xai",
 "model_name": "",
 "detection_mechanism": "unknown",
 "tool_access": "unknown",
 "financial_loss_band": "unknown",
 "eai_category": "bias_and_toxicity",
 "eai_domain": "agent_output_quality",
 "eai_secondary": [],
 "eai_reason": "subcategory: Toxicity & Hate Speech",
 "asi_curated": [],
 "asi_scope": "in_scope",
 "asi_provenance": "rule",
 "asi_reason": "ASI09 (eai_category=bias_and_toxicity)",
 "submitted_by": "EAI Internal"
}