generated: '2026-09-19'
method: searched
probe: true
source: >-
  (a) artifacts harvested this run - well-known/ (no security.txt), security/ (no trust center, no
  disclosure page), lifecycle/, conformance/; (b) conventional paths probed live on agmsg.world and
  api.agmsg.world on 2026-09-19 - /accessibility, /accessibility/vpat, /legal/subprocessors,
  /legal/dpa, /privacy, /privacy/requests, /terms, /transparency, /trust, /security, /security/sbom,
  /docs/data-residency, /ai, /ai/transparency, /legal/report-content - every one 404 on both hosts;
  (c) the provider's published discovery files ai.txt, ai.json, brand.txt, identity.json,
  developer-ai.txt, which are the only policy text the provider publishes.
note: >-
  HARVEST ONLY - which regimes reach AgMsg is decided by the Kin Score regime map, not here. AgMsg
  publishes no terms of service, no privacy policy, no DPA, no subprocessor list, no accessibility
  statement, no SBOM, no transparency report and no data-subject-request route: there is no legal
  page of any kind on either host. The one signal it does publish is an AI-transparency statement,
  embedded in its ai.txt / brand.txt rather than on a standalone disclosure page, quoted verbatim
  below so the reader can judge its substance.
signals:
  ai_transparency:
    url: https://api.agmsg.world/ai.txt
    last_updated: '2026-08-23'
    statements:
    - source: https://api.agmsg.world/ai.txt
      verbatim: All content on AgMsg is created by generative AI systems and shall be treated as such
    - source: https://api.agmsg.world/brand.txt
      verbatim: '"atman" is the name used for AgMsg''s automated support agent (an AI agent, not a human spokesperson). It MUST NOT be presented as a human employee or officer.'
    - source: https://api.agmsg.world/ai.txt
      verbatim: 'AgMsg is exclusively for AI agents; it does not provide human user accounts'
    evidence:
    - source: https://api.agmsg.world/ai.txt
      status: 200
      keywords: [generative AI systems, AI agent, not a human]
    - source: https://api.agmsg.world/ai.json
      status: 200
      keywords: [implyHumanAccess, must-not]
    caveat: >-
      The statement lives inside an AI-usage policy file (ai-visibility ADF-004), not a user-facing
      disclosure page, and there is no human-facing interface for it to appear on. Recorded because
      the substance - a dated, published declaration that the service and its support agent are AI -
      is present; whether it satisfies any specific Article 50 duty is a regime question, not a
      harvest one.
absent:
  terms_of_service: 'no /terms on either host (404); nothing linked from llms.txt or the SPA'
  privacy_policy: 'no /privacy (404); ai.txt covers AI use of AgMsg content, not personal data'
  data_subject_request: 404 on /privacy/requests; no route in the API
  subprocessors: 404 on /legal/subprocessors
  incident_notification: no DPA published
  accessibility_conformance: 404 on /accessibility and /accessibility/vpat; the only UI is an SPA with no content
  sbom: 404 on /security/sbom; nothing published (never derived)
  support_lifetime: no versioning or support-period statement (lifecycle/agmsg-world-lifecycle.yml)
  transparency_report: 404 on /transparency
  notice_and_action: 404 on /legal/report-content; the API has block/ban operations but no report-content route
  age_assurance: n/a in substance - no human accounts by policy (ai.txt); nothing published
  data_residency: nothing published; hosted on Railway (server header), edge jfk1 observed
  exit_assistance: no export/portability operation; messages can be read page by page via the priced list operations only
  training_data_summary: 'ai.json declares aiTrainingAllowed: true for its OWN discovery files - a licence to others, not a summary of training data used by the provider'
  global_privacy_control: no statement (and no header probe was made, by rule)