According to the abstract of a new arXiv preprint, the spread of autonomous AI agents in decentralized ecosystems is outpacing existing identity frameworks. The authors argue that current identity paradigms are designed for humans, who are anchored by biological traits, self-asserted attributes, and government-issued credentials. Autonomous agents have no such anchors, leaving questions about who or what they are and who is responsible for their actions.
The paper's title frames the issue as "Who is the 'I' in AI?" and its proposed direction is to establish decentralized identity and accountability for autonomous agents. The abstract does not provide implementation details, but it positions decentralized identity as a necessary response to the growing role of agents in decentralized ecosystems.
Because this digest is based on a single preprint, there are no independent sources to compare; the claims should be read as the authors' own framing rather than settled findings.