Euthyne research
Research from Euthyne on accountability for AI agents: verifiable records of what an agent did, on whose behalf, and under which permissions.
Papers
What we work on
Verifiable records of agent actions: signed receipts, their envelopes, and what they can and cannot prove.
Transparency logs and independent timestamping for agent activity.
Provenance of agent-made code changes, from agent sessions to merged changes, across harnesses and git hosts.
Privacy-preserving audit: commitments and selective disclosure, so evidence holds without exposing content.
Evidence for auditors: how agent activity maps to existing controls such as SOC 2 change management.
Each paper is published with its sources at github.com/euthyne-labs/research-euthyne. Questions or review offers: hello@euthyne.com.