EU AI Act
Requires high-risk AI systems to be traceable, accountable and subject to human oversight — yet leaves open how an autonomous agent proves who it is at each action.
Live proof-of-architecture testnet
Independent trust infrastructure for autonomous AI agents
An independent verification layer of trust for autonomous AI agents — sitting outside any single agent or platform. It is live today as a proof-of-architecture testnet, attesting agent identity and trust at the level of every action.
Global Trust Index
2.8/ 100
Evidence-weighted trust index across principal-level entities · Experimental
10,235
Monitored
45,335
Total Agreements
43,663
Settled
12
Principal-level entities
The regulatory–technical gap
Autonomous AI agents already transact, make decisions, and access systems. The AI Act, DORA, eIDAS 2.0 and AMLR require these agents to be identified, accountable, and to carry verifiable trust at the level of every action. Yet none of these regulations provides the infrastructure to implement that independently and machine-verifiably.
DORA
Demands operational resilience and verifiable oversight of digital actors across the financial sector, including the automated systems acting on its behalf.
eIDAS 2.0
Frames identity and trust services for the digital economy, but does not yet extend verifiable identity to the autonomous agents now transacting within it.
AMLR
Calls for traceability and accountability behind every transacting party — a standard autonomous agents must meet as they move value on their own.
An independent layer at the level of every action
- Step 1
Identity
Verified agent identity
Identity of the agent, separated from the agent itself and from the platform it runs on.
- Step 2
Trust
Accumulated, portable trust
Trust as a property of the agent — accumulated over time and portable across contexts.
- Step 3
Attestation
Machine-verifiable record of action
An independently audited, machine-verifiable record of every agent action.
One independently verified layer that any party can check — without having to trust the agent or its operator.
Why now
The regulatory direction is shifting the same way at once. The AI Act, DORA, eIDAS 2.0 and AMLR each, in their own way, require identified, accountable, and verifiably trusted participants. Autonomous agents are already operating inside these systems, and an independent, machine-verifiable layer of trust is becoming a requirement, not an option.
- EU AI Act
- DORA
- eIDAS 2.0
- AMLR
See the live observatory
Real-time trust signals, entities, agreements and research findings from the proof-of-architecture testnet.
View full observatory dashboard →