Delegated Authority Grants (Bounded Task Grants)
A security primitive giving an agent a short-lived, task-scoped authorization derived from (but distinct from) a human's identity — bundling delegation, policy, auditability, context, and time boundaries into one grant — replacing durable API keys/tokens as the target access model for agentic systems.
Also known as: bounded task grant, on-behalf-of grant, delegation grant, token exchange (RFC 8693), ID-JAG, scoped task grant
Part of: Agent Identity & Attribution
How authors use it
Oldest first (extraction date when no publication date is known).
| Author | Their term | Their meaning |
|---|---|---|
| Podcast: Securing AI Agents: Identity, Authorization, and the DPACT Framework Sahil Agarwal · 2026-09-23 | Bounded task grant | The target access primitive: a narrow, time-boxed, task-scoped authorization (bundling delegation + policy + context + time) that replaces a durable API key or long-lived token as 'the next platform perimeter.' |
Documents (1)
- Podcast: Securing AI Agents: Identity, Authorization, and the DPACT Framework Sahil Agarwal · 2026-09-23 source ↗ Reader 🖥️
As AI agents shift from passive chat interfaces to autonomous 'delegated actors' that take real, consequential actions with sensitive data, organizations must replace broad, human-shaped, long-lived tokens with the DPACT framework — Delegation, Policy, Auditability, Context, and Time — so that every agent acts strictly on behalf of, and never as an impersonation of, an accountable human principal, within a bounded, auditable, expiring grant of authority, adopted incrementally rather than as a big upfront program.
Propositions touching this concept (2)
By publication month, oldest first (extraction month when no publication date is known).
2026-07
- explicitThe flow uses an Identity Assertion JWT Authorisation Grant (ID-JAG) exchanged for an access token by the MCP server's authorization server, an architecture that separates identity policy from the tool call itself.
2026-09
- explicitThe industry's access-model trajectory should move from 'give the agent a key' (a durable credential) to 'issue the agent a bounded task grant' (delegation + policy + context + time bundled into one scoped, expiring authorization), positioned as the next platform perimeter replacing long-lived API keys.
Open gaps (1)
- uninformedDPACT is not benchmarked or validated against measurable outcomes. Agarwal cites two vivid but small anecdotes (a hiring bot, a food-ordering bot) and a self-applied, informal 'scoring rubric' against unnamed 'famous GitHub repos,' but offers no data on incident rates, breach frequency, or cost with versus without DPACT-style controls.Podcast: Securing AI Agents: Identity, Authorization, and the DPACT Framework — read next: extract-token-handling-ai-agents, whose cited empirical stats (88% of basic web-app attacks involve stolen credentials; one in three AI adopters already breached) would be a natural quantitative complement to DPACT's qualitative case.