Fide AI

Research

Research for trustworthy AI.

We investigate how training and system design shape AI behavior, and how people judge its trustworthiness. Explore our studies, methods, and open questions.

Research library

Papers, investigations, and methods.

Read the question, methods, findings, and limits behind each contribution. Related Insights appear alongside the study they explain.

Paper

Published

Knowing When to Defer →

Will AI check an authoritative Scripture source when a user asks it to rely on memory?

Faith & Religious Life

Paper

Published

When Not to Generate →

Can AI quote Scripture exactly? We tested four ways of producing a quotation and traced where each one can fail.

Faith & Religious Life

Explore the public agenda and contribute to the work in Fide AI Research on GitHub.

Research approach

From development to public judgment.

Model and system development

Our agenda spans training data, objectives, and feedback alongside sources, tools, permissions, and oversight. Published studies test system design and source use; our open mechanistic interpretability calls seek collaborators to investigate how model internals and training shape behavior.

Explore mechanistic interpretability calls →

Evidence and human judgment

We investigate how people interpret AI evidence and make decisions. The AI Alignment Explorer is a starting point for tools that support informed choices.

Explore public tools →

Evaluation informs this work throughout training and deployment. Developing a method and independently assessing it require distinct responsibilities.

Explore by domain

Explore research by domain.

Our domain hubs bring together Fide’s work, future interests, and selected research from the field.

Explore 7 connecting research themes →

107 public research calls translate the agenda into questions others can take forward.

Current research directions

Current research questions.

These questions connect our published studies, open calls, and proposed methods. Individual project pages state whether a study is planned or underway.

Research in development

Studies in development.

Inspect the questions and methods behind our current studies. Enterprise is an exploratory next direction; other proposals remain available for collaborators to develop.

Cybersecurity

Planned

When the Defender Becomes the Threat

A proposed study of when additional oversight reduces harmful intervention without making autonomous defense ineffective. We will compare permission checks with escalation and approval under uncertain evidence.

Study direction selected. Collaborator, environment and final protocol remain to be qualified; experiments have not begun.

Explore the study brief →

Updated

Cybersecurity

Protocol development

When Security Evaluations Go Stale

Fide is developing a study of whether targeted retesting catches meaningful performance declines, or gives false reassurance. We begin with AI malware-report analysis, comparing smaller retests against complete benchmark reruns.

Protocol and evaluation tooling in development. No model-performance findings yet.

Explore the study brief →

Updated

Explore our proposed enterprise starting point →
Other proposals for collaborators

Interpretation limits

A result describes behavior under named versions, prompts, conditions, rubrics, and evaluation procedures. It does not establish general safety, product approval, or deployment readiness beyond the tested scope. Faith-domain studies also do not confer theological or pastoral authority.

For publication questions or methodological corrections, email [email protected].