Qatar
Qatar Central Bank AI Guideline
Governance, explainability, and oversight expectations for AI used by supervised institutions.
Article 50 transparency obligations are now in active legal effect across the European Union. High-risk obligations under Annex III become enforceable in December 2027 following the Digital Omnibus agreement.
| Mandate & article | Specific runtime control |
|---|---|
| Article 12, record-keeping | Automatic logging of every agent action, with the input, rule evaluated, verdict, and timestamp written to a hash-chained record. |
| Article 14, human oversight | Rules can require approval, holding the action as PENDING_APPROVAL until a named reviewer decides. The decision is recorded against their identity. |
| Article 50, transparency | Agent interactions and their governance outcomes are recorded and exportable, so disclosure obligations can be evidenced rather than asserted. |
| Traceability over the lifecycle | Rule changes are written to a separate rule audit log, so the policy in force at any past moment can be reconstructed. |
| Evidence on demand | Compliance reports export mapped to the article they satisfy, including hash chain verification status. |
Organizations deploying public or customer-facing AI agents are legally bound by these disclosure and traceability rules today. The Act applies to providers and deployers whose AI output is used within the European Union, regardless of where the organisation is established.
Penalties reach up to 35 million euro or 7 percent of total worldwide annual turnover for prohibited practices, and up to 15 million euro or 3 percent for other breaches, whichever is higher.
Governance, explainability, and oversight expectations for AI used by supervised institutions.
Requires that machine learning or AI systems are not a black box and are capable of being audited.
Essential Cybersecurity Controls in force, with AI-specific guidance in draft.
Obligations covering autonomous and semi-autonomous systems processing personal data.
Applying the EU AI Act Article 12 Pack configures logging, retention, and export formatting against the record-keeping obligations, and turns on human approval for the rule classes most likely to be treated as high-risk. Evidence is produced as a by-product of enforcement rather than assembled afterwards.