Project summary

I've spent much of this year working on a problem in autonomous AI security: what happens when every action is individually authorized yet the sequence of actions violates a security constraint?

Our answer is a mechanism called Residual Execution Contracts. We've developed the formal foundations and tested them across more than 375,000 bounded execution extensions. The paper has been accepted at IWSEC 2026 in Tokyo.

I'm trying to get to Japan to present the work and discuss it with researchers there.

What are this project's goals? How will you achieve them?

I'll be speaking at IWSEC with people who work on security, verification and cyber-physical systems including researchers at Japanese institutions.

I'd like to come back with a clearer understanding of what our model misses and a concrete set of experiments to pursue. I'll write up the main findings within a month.

How will this funding be used?

I'm raising $2,500 to cover the trip:

• Flights: $1,250

• Conference registration: $541

• Accommodation: $500

• Local transport: $125

• Other expenses: $84

Who is on your team? What's your track record?

I work on formal verification and the security of autonomous systems.

In an earlier paper at ACM SaT-CPS 2026, we studied how to check individual AI actions before execution. The IWSEC paper grew out of a limitation of that approach: checking each step tells us surprisingly little about the sequence as a whole.

What are the most likely causes and outcomes if this project fails?

The immediate issue is money. I have an accepted paper and research discussions planned, but haven't secured enough to travel.