AI safety
AI Alignment Explorer
A map of AI safety since 2015. See how AI models score on safety tests, what AI companies have promised, which incidents have happened, and what researchers have found. Every entry links to its source.
Public tools
Tools for understanding AI and its trustworthiness.
Each tool gathers the evidence on a part of AI in one place, links every claim to its source, and says plainly what the evidence does and does not show.
Explorers
See what is known, and where it comes from.
Browse the research, tests, incidents, and policies, and follow any claim back to its source.
CybersecurityPreview
Agentic Cyber Explorer
Follow how AI agents are used to attack computer systems, how they are used to defend them, and where they fail. Each claim links to its sources and shows how strong the evidence is. Updated regularly.
- Key questions: short answers on what is known, and how sure we are
- Findings: specific claims, with the evidence behind each one
- Open questions: what researchers still need to find out
Preview: every entry has been checked against its sources, and our editors are still reviewing them. Each page says whether it has been reviewed.
Practice · Cybersecurity
Defense Evidence Lab
Learn to tell what the evidence really proves.
Work through real cybersecurity cases involving AI agents. As each new piece of evidence appears, decide what it supports, then see what it does and does not establish. It helps before you trust an AI system’s report that it stopped an attack or fixed a problem.
- Practice cases: make a call at each step as the facts come in
- Case files: real incidents, with every claim traced to its source
- Test results in context: what a published test can and cannot tell you about real use
It pairs with the Agentic Cyber Explorer: the explorer shows what has happened, and the Lab helps you judge it.
Open the Defense Evidence Lab ↗Why we build these tools
Deciding whether to use AI takes more than knowing what it can do. We want to help you answer questions like these:
- What evidence matters for what I want to do?
- What should I check before I rely on an AI answer or let it act for me?
- Which decisions should stay with people?
- When would something other than AI serve me better?
Our tools give you evidence, not verdicts. The decision stays with you, including the choice not to use AI.
More from Fide AI
Our research
Read Fide’s studies, including what we tested, how, and the limits of each result.
Insights
Plain-language articles on what AI research means for everyday and institutional decisions, with links to the evidence.
Guides by field
Research and open questions for fields such as faith, cybersecurity, and health, with finished work kept apart from plans.
Suggest a tool
Is there a decision about AI you find hard to make, at work, at home, or in your community? Tell us where clearer evidence or a practical tool would help.
Share an idea →