David Grice

AI security researcher. Creator of AgentLock, an open-source reference implementation for pre-action AI agent authorization. Building tools for adversarial testing and provenance-based agent security.

United States Joined on Mar 21, 2026 https://agentlock.dev

Education

B.S. Cybersecurity & Information Assurance

Work

AI Security Researcher, Creator of AgentLock

David Grice

Oct 29

Where agent breaches actually start: 385 incidents, eight layers, one surprise

#security #ai #agents #data

5 min read

Want to connect with David Grice?

Create an account to connect with David Grice. You can also sign in below to proceed if you already have an account.

Create Account

Already have an account? Sign in

David Grice

Jul 22

I Catalog Every AI Agent Attack. They're All the Same Bug.

#ai #agents #python #security

4 min read

David Grice

Apr 7

Why AI Agent Authorization Is Still Unsolved in 2026

#agents #ai #cybersecurity #security

7 min read

David Grice

Mar 31

How We Tripled an AI Agent's Security Score Without Changing the Model

#security #ai #python #opensource

4 min read

David Grice

Mar 21

AI Agent Tools Have No Permission Model. Here's an Open Standard to Fix It.

#ai #security #python #opensource

5 min read

DEV Community

David Grice

Badges

Writing Debut

Skills/Languages

Currently learning

Currently hacking on

Available for

Where agent breaches actually start: 385 incidents, eight layers, one surprise

Want to connect with David Grice?

I Catalog Every AI Agent Attack. They're All the Same Bug.

Why AI Agent Authorization Is Still Unsolved in 2026

How We Tripled an AI Agent's Security Score Without Changing the Model

AI Agent Tools Have No Permission Model. Here's an Open Standard to Fix It.