Identity and audit: who did what, on whose authority
Every person, AI agent, model and device gets a verifiable identity and explicit permissions — and every significant action leaves an audit record.
Provenance, identity, permission, audit
- 1
Data provenance
Where data came from, when, which version, and under what licence.
- 2
Identity
Which person, agent, model or device is acting — verified, not assumed.
- 3
Permission
What that identity is authorised to read, decide or execute.
- 4
Audit
A tamper-evident record of inputs, decisions, actions and outcomes.
What gets an identity
Users & approvers
Role-based access and approval rights.
Models & agents
Versioned models and agents with scoped permissions.
Devices
Drones, robots and edge devices bound to an identity.
Datasets
Records with provenance, licence and version.
Where it is used
- Demonstrating responsible AI practice to customers and regulators
- Investigating incidents in automated operations
- Proving which model version produced a decision
- Authorising drone missions and robot tasks
Boundaries
- Audit records support accountability; they do not by themselves guarantee correctness.
Identity & Audit — common questions
What is Trusted AI?
Trusted AI is AI whose data sources, model versions, permissions and actions can be verified, traced and audited. At DNY it is delivered through a Trusted Data Space and verifiable infrastructure rather than through claims of accuracy.
What is a Trusted Data Space?
A Trusted Data Space is a governed environment where data is shared under recorded provenance, versions, licences and permissions, so organisations can collaborate on data without giving up control of it.
Does DNY use blockchain?
Hashing, digital signatures, trusted timestamps and distributed ledgers are technology options for verification where they add value. They are tools inside the trust layer, not the identity of the platform.