Skip to content
MeetupOnlinePast event

Policy-Checked, Attributable, Revocable, Provable AI Agent Authorization

Explore the topics

Event artwork for Policy-Checked, Attributable, Revocable, Provable AI Agent Authorization

This event has ended

This listing is kept as an archive. Explore upcoming events to find your next one.

See upcoming events in Edinburgh

Supporting this event

Sponsors

Blockchain Technology Partners
About this event

What to expect

Explore secure AI agent authorization with the Caracal team in this online LFDT Scotland talk on 2 October 2026 at 3:00pm, covering policy controls, attribution and auditability.

As autonomous AI agents move into production, long-lived API keys and overly broad permissions create serious operational and security risks. A prompt injection, leaked credential or runaway agent can trigger actions without a clear answer to two essential questions: which agent acted, and under whose authority? This session examines how teams can replace credential-centric access with authorization designed specifically for agentic systems.

What the talk will cover

The Caracal team will introduce Caracal, an open-source AI authorization system being developed through the LFDT LFX Mentorship programme, with support from GitHub, Microsoft and Vercel. The presentation will explore:

  • Why conventional API keys and identity models are insufficient for AI agents
  • What “authority” means within autonomous AI systems
  • The architectural design behind Caracal
  • Policy-based authorization for individual agent actions
  • Attribution, revocation and auditability
  • Practical approaches to securing agents in production

The discussion will also explain the limitations of existing infrastructure. Identity providers can register an agent without authorizing every action; secret managers distribute credentials without restricting each task; and API gateways route requests without necessarily deciding whether an agent should be allowed to make them. Caracal’s architecture is intended to address that gap while creating evidence security teams and auditors can use to understand what happened.

Speakers and audience

The meetup will be presented by Caracal team members Ryan Madhuwala and Shubham Sharma. LFX mentees Mohammad Ali Haider and Ashutosh Singh will also share their work on the project.

This technical session is particularly relevant to people working with AI security, identity and access management, application security, open-source infrastructure, decentralized systems or production AI agents. It should also be useful for engineering and security teams evaluating how to deploy autonomous systems with narrower permissions and stronger accountability.

Format and access…

Register for this event

Create With Conference · Brighton

The UK's leading event for AI builders.

600+ builders, honest implementation stories and hands-on workshops, once a year. The next edition is being planned.

The main stage at Create With 2026 with a full room
Hackathon participants with their laptops
A panel on the main stage

Running an AI event in Edinburgh?

List it here, free.

Meetups, workshops and conferences go in front of thousands of builders after a quick review.

List your event