AI agents can initiate recommendations, decisions, communications, transactions, and multi-step workflows across enterprise systems. As autonomy increases, organizations must establish who is accountable for agent configuration, delegated authority, business outcomes, and failures. They must also retain reliable digital evidence demonstrating what the agent knew, what it attempted, which controls were applied, and who approved critical actions. This paper proposes an AI Agent Accountability and Digital Evidence Management Framework that combines ownership models, decision rights, agent identity, delegation records, audit trails, evidence integrity, retention, legal hold, access control, chain of custody, and investigation workflows. The framework defines accountability across business owners, agent owners, model owners, data owners, platform teams, and human approvers. It also addresses evidence from prompts, retrieved context, tool calls, policy decisions, approvals, outputs, exceptions, and downstream actions. The proposed approach enables enterprises to operate AI agents with stronger governance, audit readiness, incident response, and defensible accountability.

Sanjeeve Kumar Gajadi, "AI Agent Accountability and Digital Evidence Management Framework" International Journal of Scientific Research and Engineering Development, V9(5): Page(1857-1858) September - October 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.