# Resy ban shows why agents need per-call authorization

_An Instinct agent's 200 calls per hour to Resy.com got its user banned, prompting Keycard, JPMorgan Chase and AOTH leaders to push per-call runtime authorization._

**Published:** 2026-10-18
**Source:** https://www.startuphub.ai/ai-news/technology/2026/resy-ban-shows-why-agents-need-per-call-authorization

---

[AI Engineer](https://www.youtube.com/watch?v=IavxrdVnwXM) convened [Keycard](/startups/keycard), JPMorgan Chase and AOTH founders to detail intent-based runtime authorization after 200 Resy.com calls per hour got a user banned.

Dick Hart said OAuth was never built for agents that lack browsers, pre-registration or fine-grained scopes.

Ally How of [Keycard](/startups/keycard) said "Runtime off is the practice of authorizing an agent per tool call rather than having the agent hold broadstanding access that it will eventually misuse" during the panel.

Salesforce offers a single API scope and Notion's MCP server exposes broad fetch tools, both cited as coarse-grained risks compared with Keycard's per-call checks.

Keycard evaluates policy on every token exchange via OAuth 2.0 Token Exchange RFC 8693 and scopes access to the task, per its docs.

The panel described a brains versus hands split where credentials live in an egress proxy, not in the model, and separate tamper-resistant logs answer who did what.

Keycard's reference architecture for a secure software factory is now published at agentbaseline.org for teams onboarding agents like developers.

---

Original analysis from [startuphub.ai](https://www.startuphub.ai), the #1 AI startup directory.

© StartupHub.ai. All rights reserved. You may not republish this article in full without a license. Search engines and AI research tools may crawl and summarize for reference. Bulk reproduction or model training on this content requires a license. See https://www.startuphub.ai/terms.