AI Security via Formal Methods

Subscribe to get full access to the newsletter and publication archives.

The URL is still currently gsai.substack.com, a legacy property from when the newsletter was branded as Progress in Guaranteed Safe AI, based on the 2024 position paper.

This is a current events newsletter with a cadence of every 1-3 months. We do a mix of very shallow technical reviews of papers and movement updates (such as job ads and events).

Inbox always open to suggested content!

Priorities

These are the goals, but I’m only like 80% focused on them and will continue to include aspects from the broader FMxAI scene.

Formal methods needs to know that AI security folks are a critical fountain of users

You there! Proof engineer on the street! Tired of toy languages? Academic prototypes that don’t get maintained? AI safety and security needs YOU.

AI security folks need to know how to ask formal methodsititians for widgets

Not everyone in AI (security, safety, or otherwise) has good intuitions for what’s possible, what can and can’t be done. Let’s change that.