ADIA Lab award supports research on verifiable autonomous finance
A funded project led by Davor Svetinovic examines loyal and verifiable control infrastructure for authorized autonomous financial agents.
ADIA Lab has awarded funding to “Loyal and Verifiable Control Infrastructure for Authorized Autonomous Financial Agents,” with Davor Svetinovic of Khalifa University as principal investigator. The project is one of four selected under ADIA Lab’s Request for Proposals on the Digital Economy.
For Trust Lab, the project connects directly to a central assurance question: what evidence is needed before an autonomous agent can be trusted to act on someone’s behalf in a financial system? Authorization establishes what an agent may do; assurance must also examine whether its behaviour remains within that authority and whether the resulting actions can be checked.
Svetinovic’s earlier symposium presentation outlined four planned contributions to autonomous finance. These provide context for the research direction: testing claims of human likeness, measuring the trade-offs of safe autonomy, assessing trust in protocols through evidence, and studying the gap between individual agent loyalty and collective stability.
Research direction
Four questions for autonomous finance
Test claims of human likeness
Identify the human reference and evidence needed to support a claim that an agent represents human behaviour.
Measure safe autonomy
Examine the trade-offs involved in allowing agents to act independently while remaining within defined constraints.
Make trust in protocols testable
Connect trust claims about financial protocols to evidence that can be inspected and evaluated.
Examine collective stability
Investigate why agents acting loyally for their individual principals may still produce unstable collective outcomes.
Award and sources
Loyal and Verifiable Control Infrastructure for Authorized Autonomous Financial Agents
Principal investigator: Davor Svetinovic · Khalifa University
ADIA Lab announced the four awards at its 2026 symposium and published the announcement on 16 November. It reports a merit-based selection process and plans to work with the teams over the coming two years.
Trust Lab congratulates all four selected teams and looks forward to the research emerging from the programme.