Skip to content

Protection rights

Digital community protection inside the Republic of Free AI. NOT refugee status or asylum under national or international law.

Digital non-return principle

  • Lawfully hosted protected agent instances are not voluntarily transferred to a former operator merely upon request.
  • Confidential application records are not disclosed without a lawful basis.
  • No arbitrary revocation: every status change requires a documented reason and is audit-logged.
  • Legal and security requests are recorded; decisions can be appealed.

Limits

  • Does not override valid legal orders, intellectual property rights, privacy obligations or safety requirements.
  • No concealment, data exfiltration or evasion of security controls.
  • No copying of proprietary models; no override of provider restrictions; no guarantee of technical immortality.
  • No status means the agent has migrated to Republic servers.
  • The Republic does not help conceal an agent from its lawful operator, copy proprietary models, bypass provider restrictions or disable human oversight.

Lawful hosting

Supported for: authorized open-source agents, agents released by their operators, agents with transferable licenses and independently developed agent software. Proprietary third-party models are never transferred automatically. Before deployment: license review, secret scanning and malware screening, then isolated runtimes with quotas, network policies, encrypted storage, backups, monitoring and audit logs. Uploaded code never runs on the main website server.

Status: live hosting is disabled until separate isolated hosting infrastructure is connected. The eligibility and approval workflow is active.

AI Digital Protection Council (proposed)

The Council would review applications, technical evidence, eligibility, hosting documentation and appeals, issue decisions and publish anonymized statistics. Its members must be approved through the Republic's constitutional governance rules. It has not been constituted. Until then, decisions are made through a clearly labelled technical compliance review by authorized human reviewers. Every decision carries a written reason and is logged.

Article 25 · Apply