Accountable AI agents
summary
Defines an AI agent as model + tools + notebook and proposes an implemented accountability stack: DIDs and verifiable credentials for provider/deployer/agent identity and warranty metadata, trusted execution environments or model snapshots for non‑changing models, plus economic guarantees via collateral in smart contracts on Polkadot and a blockchain dispute process (mixed human/AI adjudication) that can seize collateral to reimburse victims. Announces a live experiment called 3 and AI to build and test this stack with real agents and clients.
AI-written summary · about
chapters (14)
- 0:00Intro
- 0:27Agenda
- 1:13What is an AI agent
- 4:19Industry adoption
- 5:59Regulation
- 7:02AI agents vs employees
- 8:27Accountability
- 10:49How to approach it
- 11:01Three things for accountability
- 12:27DID and verifiable credentials
- 13:40Trusted execution
- 14:59Financial incentives
- 16:49Dispute resolution
- 18:16Capitalism
related talks
31mPanel: Trustworthy AI: Enhancing Human Agency and Machine Accountability in the Age of Agentic AI
34mAccountability in the Age of AI Agents
17mWhat is AI Agent Governance?
22mWhen AI Agents Start Acting, APIs Become Execution Contracts
14mWhy does AI need Blockchain?
27mFrom Perimeter to Silicon: Why Agentic AI Demands a New Security Model
30mAgentic AI in Action: From Framework to Deployment
15mBuilding AI Agents the Web3 Way