Editorial Policy

The Institute studies how artificial intelligence changes the balance between cyber offense and cyber defense. A study is taken on when its question can be answered by measurement, when the answer would change a decision a security team makes, and when a method and a primary endpoint can be fixed before any data exists.

Each study’s protocol is posted before data collection begins. After that point the protocol changes only by a dated amendment published beside the original. Registered studies are listed on the Studies page.

Open Science Framework records will be linked here, and on each study’s page, as each study is registered there. None is posted yet.

A figure derived from client data is published only when it rests on an aggregate n of at least 20 and, for any segment reported, an n of at least 10. No table cell reports an n below 5.

Claims are sourced to primary texts and cited by section, so a reader can find the exact passage relied on. A DOI is given wherever one exists. No paper is described as peer-reviewed unless a venue has accepted it.

An error found after publication is corrected on the page where it appeared, with a dated correction note that states what changed. Nothing is edited silently.

The Institute is a research division of Bridgham Cyber, LLC, which funds and staffs it. That relationship is stated here, on the Institute Facts page, and in each publication, together with any other interest bearing on that publication. The Statement on Research Ethics and Integrity sets out the commitments that bind this work.

Every publication is free to read. Copyright Bridgham Cyber, LLC. All rights reserved.

Questions about this policy, and reports of possible errors, go to headofresearch@bridghamcyber.com.