Paper proposes agentic attribution of cloud resource ownership
Original title: Agentic Attribution of Cloud Resource Ownership
On 100 AWS resources the architecture reached 92.1% weighted recall and 86.6% weighted precision, versus 54.8% and 77.4%.
A cs.CR paper defines cloud resource ownership as people or teams linked by five strength-ordered patterns and presents an agentic architecture that derives ownership from work records, code history, audit logs and identity directories. An offline profile produces tested signal procedures and a playbook; a coordinator and parallel investigators then attribute each resource, with rule-based confidence. On 100 resources across 50 AWS types in one organization, candidates were pooled with an unprimed agent and evidence was checked against records. The architecture cost 0.55 times the agent's cost per unit of coverage.
Unconfirmed: Results are from one organization and were verified by the authors against internal records.
- Evaluation covers 100 resources of 50 AWS resource types in one organization.
- Weighted recall is 92.1% and weighted precision 86.6%.
- The unprimed agent scored 54.8% recall and 77.4% precision.
- Gains come from builders found in work records and code history.
From the original article
1 min read at arxiv.orgSecurity tools raise findings across cloud environments, and AI agents now investigate and prioritize them. A finding is still resolved by people: it has to reach the people responsible for fixing it and accountable for the resource it concerns. Ownership is the organizational context that decides who that is.
By Tiltan Gilat · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.